2013-05-28 05:54:09 +00:00
|
|
|
/*
|
|
|
|
* s3fs - FUSE-based file system backed by Amazon S3
|
|
|
|
*
|
|
|
|
* Copyright 2007-2013 Takeshi Nakatani <ggtakec.com>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
* as published by the Free Software Foundation; either version 2
|
|
|
|
* of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#include <sys/types.h>
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include <sys/time.h>
|
|
|
|
#include <sys/file.h>
|
2013-08-19 06:29:24 +00:00
|
|
|
#include <stdint.h>
|
2013-05-28 05:54:09 +00:00
|
|
|
#include <unistd.h>
|
|
|
|
#include <pthread.h>
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include <syslog.h>
|
|
|
|
#include <errno.h>
|
2013-05-28 05:54:09 +00:00
|
|
|
#include <string.h>
|
|
|
|
#include <assert.h>
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include <curl/curl.h>
|
|
|
|
#include <openssl/crypto.h>
|
2013-05-28 05:54:09 +00:00
|
|
|
#include <string>
|
|
|
|
#include <iostream>
|
|
|
|
#include <sstream>
|
|
|
|
#include <map>
|
2013-06-21 06:07:22 +00:00
|
|
|
#include <list>
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include <vector>
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include "common.h"
|
2013-05-28 05:54:09 +00:00
|
|
|
#include "fdcache.h"
|
|
|
|
#include "s3fs.h"
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include "s3fs_util.h"
|
2013-11-17 08:50:41 +00:00
|
|
|
#include "string_util.h"
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
#include "curl.h"
|
2013-05-28 05:54:09 +00:00
|
|
|
|
|
|
|
using namespace std;
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
//------------------------------------------------
|
|
|
|
// Symbols
|
|
|
|
//------------------------------------------------
|
|
|
|
#define MAX_OBJECT_SIZE 68719476735LL // 64GB - 1L
|
|
|
|
#define MULTIPART_LOWLIMIT (20 * 1024 * 1024) // 20MB
|
|
|
|
#define FDPAGE_SIZE (50 * 1024 * 1024) // 50MB(parallel uploading is 5 parallel(default) * 10 MB)
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
//------------------------------------------------
|
|
|
|
// CacheFileStat class methods
|
|
|
|
//------------------------------------------------
|
|
|
|
bool CacheFileStat::MakeCacheFileStatPath(const char* path, string& sfile_path, bool is_create_dir)
|
2013-06-21 06:07:22 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// make stat dir top path( "/<cache_dir>/.<bucket_name>.stat" )
|
|
|
|
string top_path = FdManager::GetCacheDir();
|
|
|
|
top_path += "/.";
|
|
|
|
top_path += bucket;
|
|
|
|
top_path += ".stat";
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(is_create_dir){
|
|
|
|
mkdirp(top_path + mydirname(path), 0777);
|
|
|
|
}
|
|
|
|
if(!path || '\0' == path[0]){
|
|
|
|
sfile_path = top_path;
|
|
|
|
}else{
|
|
|
|
sfile_path = top_path + SAFESTRPTR(path);
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2013-07-29 08:20:19 +00:00
|
|
|
bool CacheFileStat::DeleteCacheFileStat(const char* path)
|
|
|
|
{
|
|
|
|
if(!path || '\0' == path[0]){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// stat path
|
|
|
|
string sfile_path;
|
|
|
|
if(!CacheFileStat::MakeCacheFileStatPath(path, sfile_path, false)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("failed to create cache stat file path(%s)", path);
|
2013-07-29 08:20:19 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(0 != unlink(sfile_path.c_str())){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("failed to delete file(%s): errno=%d", path, errno);
|
2013-07-29 08:20:19 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
//------------------------------------------------
|
|
|
|
// CacheFileStat methods
|
|
|
|
//------------------------------------------------
|
|
|
|
CacheFileStat::CacheFileStat(const char* tpath) : path(""), fd(-1)
|
|
|
|
{
|
|
|
|
if(tpath && '\0' != tpath[0]){
|
|
|
|
SetPath(tpath, true);
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
CacheFileStat::~CacheFileStat()
|
2013-06-21 06:07:22 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
Release();
|
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool CacheFileStat::SetPath(const char* tpath, bool is_open)
|
|
|
|
{
|
|
|
|
if(!tpath || '\0' == tpath[0]){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(!Release()){
|
|
|
|
// could not close old stat file.
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(tpath){
|
|
|
|
path = tpath;
|
|
|
|
}
|
|
|
|
if(!is_open){
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return Open();
|
|
|
|
}
|
|
|
|
|
|
|
|
bool CacheFileStat::Open(void)
|
|
|
|
{
|
|
|
|
if(0 == path.size()){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(-1 != fd){
|
|
|
|
// already opened
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
// stat path
|
|
|
|
string sfile_path;
|
|
|
|
if(!CacheFileStat::MakeCacheFileStatPath(path.c_str(), sfile_path, true)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to create cache stat file path(%s)", path.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// open
|
|
|
|
if(-1 == (fd = open(sfile_path.c_str(), O_CREAT|O_RDWR, 0600))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("failed to open cache stat file path(%s) - errno(%d)", path.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// lock
|
|
|
|
if(-1 == flock(fd, LOCK_EX)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to lock cache stat file(%s) - errno(%d)", path.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
close(fd);
|
|
|
|
fd = -1;
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// seek top
|
|
|
|
if(0 != lseek(fd, 0, SEEK_SET)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to lseek cache stat file(%s) - errno(%d)", path.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
flock(fd, LOCK_UN);
|
|
|
|
close(fd);
|
|
|
|
fd = -1;
|
|
|
|
return false;
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("file locked(%s - %s)", path.c_str(), sfile_path.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
return true;
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool CacheFileStat::Release(void)
|
2013-06-21 06:07:22 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 == fd){
|
|
|
|
// already release
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
// unlock
|
|
|
|
if(-1 == flock(fd, LOCK_UN)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to unlock cache stat file(%s) - errno(%d)", path.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("file unlocked(%s)", path.c_str());
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 == close(fd)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to close cache stat file(%s) - errno(%d)", path.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
fd = -1;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
//------------------------------------------------
|
|
|
|
// PageList methods
|
|
|
|
//------------------------------------------------
|
|
|
|
void PageList::FreeList(fdpage_list_t& list)
|
|
|
|
{
|
|
|
|
for(fdpage_list_t::iterator iter = list.begin(); iter != list.end(); iter = list.erase(iter)){
|
|
|
|
delete (*iter);
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
list.clear();
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
PageList::PageList(off_t size, bool is_init)
|
2013-06-21 06:07:22 +00:00
|
|
|
{
|
2013-09-14 21:50:39 +00:00
|
|
|
Init(size, is_init);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
PageList::~PageList()
|
|
|
|
{
|
|
|
|
Clear();
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t PageList::Size(void) const
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
|
|
|
if(0 == pages.size()){
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
fdpage_list_t::const_reverse_iterator riter = pages.rbegin();
|
|
|
|
return ((*riter)->offset + (*riter)->bytes);
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
int PageList::Resize(off_t size, bool is_init)
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t total = Size();
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(0 == total){
|
|
|
|
Init(size, is_init);
|
|
|
|
|
|
|
|
}else if(total < size){
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t remain = size - total; // remaining bytes
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fdpage_list_t::reverse_iterator riter = pages.rbegin();
|
|
|
|
|
|
|
|
if((*riter)->bytes < FdManager::GetPageSize()){
|
|
|
|
// resize last area
|
|
|
|
remain += (*riter)->bytes; // remaining bytes(without last page)
|
2013-11-17 08:50:41 +00:00
|
|
|
(*riter)->bytes = remain > static_cast<off_t>(FdManager::GetPageSize()) ? FdManager::GetPageSize() : static_cast<size_t>(remain); // reset page size
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
remain -= (*riter)->bytes; // remaining bytes(after last page)
|
|
|
|
(*riter)->init = is_init;
|
|
|
|
}
|
|
|
|
|
|
|
|
// add new area
|
|
|
|
for(off_t next = (*riter)->next(); 0 < remain; remain -= size, next += size){
|
2013-11-17 08:50:41 +00:00
|
|
|
size = remain > static_cast<off_t>(FdManager::GetPageSize()) ? static_cast<off_t>(FdManager::GetPageSize()) : remain;
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fdpage* page = new fdpage(next, size, is_init);
|
|
|
|
pages.push_back(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
}else if(total > size){
|
|
|
|
for(fdpage_list_t::reverse_iterator riter = pages.rbegin(); riter != pages.rend(); riter++){
|
2013-11-17 08:50:41 +00:00
|
|
|
if((*riter)->offset < size){
|
|
|
|
(*riter)->bytes = static_cast<size_t>(size - (*riter)->offset);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
break;
|
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
void PageList::Clear(void)
|
|
|
|
{
|
|
|
|
PageList::FreeList(pages);
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
int PageList::Init(off_t size, bool is_init)
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
|
|
|
Clear();
|
2013-11-17 08:50:41 +00:00
|
|
|
for(off_t total = 0; total < size; total += FdManager::GetPageSize()){
|
|
|
|
size_t areasize = (total + FdManager::GetPageSize()) < size ? FdManager::GetPageSize() : static_cast<size_t>(size - total);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fdpage* page = new fdpage(total, areasize, is_init);
|
|
|
|
pages.push_back(page);
|
|
|
|
}
|
|
|
|
return pages.size();
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
bool PageList::IsInit(off_t start, off_t size)
|
2013-06-21 06:07:22 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
off_t next = start + size;
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(0 == pages.size()){
|
2013-06-21 06:07:22 +00:00
|
|
|
return false;
|
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// check end
|
|
|
|
fdpage_list_t::reverse_iterator riter = pages.rbegin();
|
|
|
|
if((*riter)->next() < next){
|
|
|
|
// size is over end of page list.
|
2013-06-21 06:07:22 +00:00
|
|
|
return false;
|
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++){
|
|
|
|
if(next <= (*iter)->offset){
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if((start <= (*iter)->offset && (*iter)->offset < next) || // start < iter-start < end
|
|
|
|
(start <= (*iter)->end() && (*iter)->end() < next) || // start < iter-end < end
|
|
|
|
((*iter)->offset <= start && next <= (*iter)->end()) ) // iter-start < start < end < iter-end
|
|
|
|
{
|
|
|
|
if(!(*iter)->init){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
bool PageList::SetInit(off_t start, off_t size, bool is_init)
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
|
|
|
// check size & resize
|
|
|
|
if(Size() < (start + size)){
|
|
|
|
Resize(start + size, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
off_t next = start + size;
|
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++){
|
|
|
|
if((*iter)->end() < start){
|
|
|
|
// out of area
|
|
|
|
// iter:start < iter:end < start < end
|
|
|
|
continue;
|
|
|
|
}else if(next <= (*iter)->offset){
|
|
|
|
// out of area
|
|
|
|
// start < end < iter:start < iter:end
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// area of target overlaps with iter area
|
|
|
|
// iter:start < start < iter:end < end
|
|
|
|
// iter:start < start < end < iter:end
|
|
|
|
// start < iter:start < iter:end < end
|
|
|
|
// start < iter:start < end < iter:end
|
|
|
|
if((*iter)->init != is_init){
|
|
|
|
(*iter)->init = is_init;
|
|
|
|
}
|
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool PageList::FindUninitPage(off_t start, off_t& resstart, size_t& ressize)
|
|
|
|
{
|
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++){
|
|
|
|
if(start <= (*iter)->end()){
|
|
|
|
if(!(*iter)->init){
|
|
|
|
resstart = (*iter)->offset;
|
|
|
|
ressize = (*iter)->bytes;
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
int PageList::GetUninitPages(fdpage_list_t& uninit_list, off_t start)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++){
|
|
|
|
if(start <= (*iter)->end()){
|
|
|
|
// after start pos
|
|
|
|
if(!(*iter)->init){
|
|
|
|
// found uninitialized area
|
|
|
|
fdpage_list_t::reverse_iterator riter = uninit_list.rbegin();
|
|
|
|
if(riter != uninit_list.rend() && (*riter)->next() == (*iter)->offset){
|
|
|
|
// merge to before page
|
|
|
|
(*riter)->bytes += (*iter)->bytes;
|
|
|
|
}else{
|
|
|
|
fdpage* page = new fdpage((*iter)->offset, (*iter)->bytes, false);
|
|
|
|
uninit_list.push_back(page);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return uninit_list.size();
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool PageList::Serialize(CacheFileStat& file, bool is_output)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(!file.Open()){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(is_output){
|
|
|
|
//
|
|
|
|
// put to file
|
|
|
|
//
|
|
|
|
stringstream ssall;
|
|
|
|
ssall << Size();
|
|
|
|
|
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++){
|
|
|
|
ssall << "\n" << (*iter)->offset << ":" << (*iter)->bytes << ":" << ((*iter)->init ? "1" : "0");
|
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
string strall = ssall.str();
|
|
|
|
if(0 >= pwrite(file.GetFd(), strall.c_str(), strall.length(), 0)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to write stats(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
2013-05-28 05:54:09 +00:00
|
|
|
}else{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
//
|
|
|
|
// loading from file
|
|
|
|
//
|
|
|
|
struct stat st;
|
|
|
|
memset(&st, 0, sizeof(struct stat));
|
|
|
|
if(-1 == fstat(file.GetFd(), &st)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("fstat is failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(0 >= st.st_size){
|
|
|
|
// nothing
|
|
|
|
Init(0, false);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
char* ptmp;
|
|
|
|
if(NULL == (ptmp = (char*)calloc(st.st_size + 1, sizeof(char)))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNCRIT("could not allocate memory.");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
S3FS_FUSE_EXIT();
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// read from file
|
|
|
|
if(0 >= pread(file.GetFd(), ptmp, st.st_size, 0)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to read stats(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
free(ptmp);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
string oneline;
|
|
|
|
stringstream ssall(ptmp);
|
|
|
|
|
|
|
|
// init
|
|
|
|
Clear();
|
|
|
|
|
|
|
|
// load(size)
|
|
|
|
if(!getline(ssall, oneline, '\n')){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to parse stats.");
|
2013-09-14 21:50:39 +00:00
|
|
|
free(ptmp);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t total = s3fs_strtoofft(oneline.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
// load each part
|
|
|
|
bool is_err = false;
|
|
|
|
while(getline(ssall, oneline, '\n')){
|
|
|
|
string part;
|
|
|
|
stringstream ssparts(oneline);
|
|
|
|
// offset
|
|
|
|
if(!getline(ssparts, part, ':')){
|
|
|
|
is_err = true;
|
|
|
|
break;
|
|
|
|
}
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t offset = s3fs_strtoofft(part.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// size
|
|
|
|
if(!getline(ssparts, part, ':')){
|
|
|
|
is_err = true;
|
|
|
|
break;
|
|
|
|
}
|
2013-11-17 08:50:41 +00:00
|
|
|
off_t size = s3fs_strtoofft(part.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// init
|
|
|
|
if(!getline(ssparts, part, ':')){
|
|
|
|
is_err = true;
|
|
|
|
break;
|
|
|
|
}
|
2013-11-17 08:50:41 +00:00
|
|
|
bool is_init = (1 == s3fs_strtoofft(part.c_str()) ? true : false);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// add new area
|
|
|
|
SetInit(offset, size, is_init);
|
|
|
|
}
|
2013-09-14 21:50:39 +00:00
|
|
|
free(ptmp);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(is_err){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to parse stats.");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
Clear();
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
// check size
|
|
|
|
if(total != Size()){
|
2013-11-17 08:50:41 +00:00
|
|
|
DPRN("different size(%jd - %jd).", (intmax_t)total, (intmax_t)Size());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
Clear();
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
void PageList::Dump(void)
|
|
|
|
{
|
|
|
|
int cnt = 0;
|
|
|
|
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("pages = {");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
for(fdpage_list_t::iterator iter = pages.begin(); iter != pages.end(); iter++, cnt++){
|
2013-08-19 06:29:24 +00:00
|
|
|
DPRNINFO(" [%08d] -> {%014jd - %014zu : %s}", cnt, (intmax_t)((*iter)->offset), (*iter)->bytes, (*iter)->init ? "true" : "false");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("}");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
//------------------------------------------------
|
|
|
|
// FdEntity methods
|
|
|
|
//------------------------------------------------
|
|
|
|
FdEntity::FdEntity(const char* tpath, const char* cpath)
|
|
|
|
: is_lock_init(false), path(SAFESTRPTR(tpath)), cachepath(SAFESTRPTR(cpath)), fd(-1), file(NULL), is_modify(false)
|
|
|
|
{
|
|
|
|
try{
|
|
|
|
pthread_mutex_init(&fdent_lock, NULL);
|
|
|
|
is_lock_init = true;
|
|
|
|
}catch(exception& e){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNCRIT("failed to init mutex");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
FdEntity::~FdEntity()
|
|
|
|
{
|
|
|
|
Clear();
|
|
|
|
|
|
|
|
if(is_lock_init){
|
|
|
|
try{
|
|
|
|
pthread_mutex_destroy(&fdent_lock);
|
|
|
|
}catch(exception& e){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNCRIT("failed to destroy mutex");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
is_lock_init = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void FdEntity::Clear(void)
|
|
|
|
{
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
|
|
|
|
if(file){
|
2013-08-22 09:36:16 +00:00
|
|
|
if(0 != cachepath.size()){
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
CacheFileStat cfstat(path.c_str());
|
|
|
|
if(!pagelist.Serialize(cfstat, true)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to save cache stat file(%s).", path.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
fd = -1;
|
|
|
|
}
|
|
|
|
pagelist.Init(0, false);
|
|
|
|
refcnt = 0;
|
|
|
|
path = "";
|
|
|
|
cachepath = "";
|
|
|
|
is_modify = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
void FdEntity::Close(void)
|
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][refcnt=%d]", path.c_str(), fd, (-1 != fd ? refcnt - 1 : refcnt));
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 != fd){
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
|
|
|
|
if(0 < refcnt){
|
|
|
|
refcnt--;
|
|
|
|
}
|
|
|
|
if(0 == refcnt){
|
2013-08-22 09:36:16 +00:00
|
|
|
if(0 != cachepath.size()){
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
CacheFileStat cfstat(path.c_str());
|
|
|
|
if(!pagelist.Serialize(cfstat, true)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to save cache stat file(%s).", path.c_str());
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
fd = -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
int FdEntity::Dup(void)
|
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][refcnt=%d]", path.c_str(), fd, (-1 != fd ? refcnt + 1 : refcnt));
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 != fd){
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
refcnt++;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return fd;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
int FdEntity::Open(off_t size, time_t time)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool already_opened = false; // already opened fd
|
|
|
|
bool is_csf_loaded = false; // loaded by cache stat file
|
|
|
|
bool is_truncate = false; // need to truncate
|
|
|
|
bool init_value = false; // value for pagelist
|
2013-05-28 05:54:09 +00:00
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][size=%jd][time=%jd]", path.c_str(), fd, (intmax_t)size, (intmax_t)time);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 != fd){
|
|
|
|
// already opened, needs to increment refcnt.
|
|
|
|
already_opened = true;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
|
|
|
}else{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// open
|
|
|
|
if(0 != cachepath.size()){
|
|
|
|
// At first, open & flock stat file.
|
|
|
|
{
|
|
|
|
CacheFileStat cfstat(path.c_str());
|
|
|
|
is_csf_loaded = pagelist.Serialize(cfstat, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
// open cache file
|
|
|
|
if(is_csf_loaded && -1 != (fd = open(cachepath.c_str(), O_RDWR))){
|
|
|
|
// file exists
|
|
|
|
struct stat st;
|
|
|
|
memset(&st, 0, sizeof(struct stat));
|
|
|
|
if(-1 == fstat(fd, &st)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("fstat is failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
fd = -1;
|
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
2013-11-17 08:50:41 +00:00
|
|
|
if((-1 != size && size != pagelist.Size()) || st.st_size != pagelist.Size()){
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
is_csf_loaded = false; // reinitializing
|
|
|
|
if(-1 == size){
|
|
|
|
size = st.st_size;
|
|
|
|
}else{
|
|
|
|
is_truncate = true;
|
|
|
|
}
|
|
|
|
}else{
|
|
|
|
// size OK! --> no initialize after this line.
|
|
|
|
}
|
|
|
|
|
|
|
|
}else{
|
|
|
|
// file does not exist -> create & open
|
|
|
|
if(-1 == (fd = open(cachepath.c_str(), O_CREAT|O_RDWR|O_TRUNC, 0600))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to open file(%s). errno(%d)", cachepath.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
|
|
|
if(-1 == size){
|
|
|
|
size = 0;
|
|
|
|
}else{
|
|
|
|
is_truncate = true;
|
|
|
|
}
|
|
|
|
is_csf_loaded = false;
|
|
|
|
}
|
|
|
|
// make file pointer(for being same tmpfile)
|
|
|
|
if(NULL == (file = fdopen(fd, "wb"))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to get fileno(%s). errno(%d)", cachepath.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
close(fd);
|
|
|
|
fd = -1;
|
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
|
|
|
|
|
|
|
}else{
|
|
|
|
// open temporary file
|
|
|
|
if(NULL == (file = tmpfile()) || -1 ==(fd = fileno(file))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to open tmp file. err(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(file){
|
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
}
|
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
|
|
|
if(-1 == size){
|
|
|
|
size = 0;
|
|
|
|
}else{
|
|
|
|
is_truncate = true;
|
|
|
|
}
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// truncate
|
|
|
|
if(is_truncate){
|
|
|
|
if(0 != ftruncate(fd, size) || 0 != fsync(fd)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("ftruncate(%s) or fsync returned err(%d)", cachepath.c_str(), errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
fd = -1;
|
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// set mtime
|
|
|
|
if(-1 != time){
|
|
|
|
if(0 != SetMtime(time)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to set mtime. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
fclose(file);
|
|
|
|
file = NULL;
|
|
|
|
fd = -1;
|
|
|
|
return (0 == errno ? -EIO : -errno);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// set internal data
|
|
|
|
if(already_opened){
|
|
|
|
Dup();
|
|
|
|
}else{
|
|
|
|
if(!is_csf_loaded){
|
|
|
|
pagelist.Init(size, init_value);
|
|
|
|
}
|
|
|
|
refcnt = 1;
|
|
|
|
is_modify = false;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int FdEntity::SetMtime(time_t time)
|
|
|
|
{
|
2013-08-19 06:29:24 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][time=%jd]", path.c_str(), fd, (intmax_t)time);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 == time){
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if(-1 != fd){
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
|
|
|
|
struct timeval tv[2];
|
|
|
|
tv[0].tv_sec = time;
|
|
|
|
tv[0].tv_usec= 0L;
|
|
|
|
tv[1].tv_sec = tv[0].tv_sec;
|
|
|
|
tv[1].tv_usec= 0L;
|
|
|
|
if(-1 == futimes(fd, tv)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("futimes failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
}else if(0 < cachepath.size()){
|
|
|
|
// not opened file yet.
|
|
|
|
struct utimbuf n_mtime;
|
|
|
|
n_mtime.modtime = time;
|
|
|
|
n_mtime.actime = time;
|
|
|
|
if(-1 == utime(cachepath.c_str(), &n_mtime)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("utime failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
bool FdEntity::GetSize(off_t& size)
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
|
|
|
if(-1 == fd){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
size = pagelist.Size();
|
2013-05-28 05:54:09 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool FdEntity::GetMtime(time_t& time)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
struct stat st;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(!GetStats(st)){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
time = st.st_mtime;
|
|
|
|
return true;
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool FdEntity::GetStats(struct stat& st)
|
|
|
|
{
|
|
|
|
if(-1 == fd){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
memset(&st, 0, sizeof(struct stat));
|
|
|
|
if(-1 == fstat(fd, &st)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("fstat failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return true;
|
|
|
|
}
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
bool FdEntity::SetAllStatus(bool is_enable)
|
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][%s]", path.c_str(), fd, is_enable ? "enable" : "disable");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 == fd){
|
|
|
|
return false;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// get file size
|
|
|
|
struct stat st;
|
|
|
|
memset(&st, 0, sizeof(struct stat));
|
|
|
|
if(-1 == fstat(fd, &st)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("fstat is failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
// Reinit
|
|
|
|
pagelist.Init(st.st_size, is_enable);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
int FdEntity::Load(off_t start, off_t size)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
int result = 0;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][offset=%jd][size=%jd]", path.c_str(), fd, (intmax_t)start, (intmax_t)size);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 == fd){
|
|
|
|
return -EBADF;
|
|
|
|
}
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// check loaded area & load
|
|
|
|
fdpage_list_t uninit_list;
|
|
|
|
if(0 < pagelist.GetUninitPages(uninit_list, start)){
|
|
|
|
for(fdpage_list_t::iterator iter = uninit_list.begin(); iter != uninit_list.end(); iter++){
|
|
|
|
if(-1 != size && (start + size) <= (*iter)->offset){
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// download
|
|
|
|
if((*iter)->bytes >= MULTIPART_LOWLIMIT && !nomultipart){ // 20MB
|
|
|
|
// parallel request
|
|
|
|
// Additional time is needed for large files
|
|
|
|
time_t backup = 0;
|
|
|
|
if(120 > S3fsCurl::GetReadwriteTimeout()){
|
|
|
|
backup = S3fsCurl::SetReadwriteTimeout(120);
|
|
|
|
}
|
|
|
|
result = S3fsCurl::ParallelGetObjectRequest(path.c_str(), fd, (*iter)->offset, (*iter)->bytes);
|
|
|
|
if(0 != backup){
|
|
|
|
S3fsCurl::SetReadwriteTimeout(backup);
|
|
|
|
}
|
|
|
|
}else{
|
|
|
|
// single request
|
|
|
|
S3fsCurl s3fscurl;
|
|
|
|
result = s3fscurl.GetObjectRequest(path.c_str(), fd, (*iter)->offset, (*iter)->bytes);
|
|
|
|
}
|
|
|
|
if(0 != result){
|
|
|
|
break;
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
// Set init flag
|
2013-11-17 08:50:41 +00:00
|
|
|
pagelist.SetInit((*iter)->offset, static_cast<off_t>((*iter)->bytes), true);
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
PageList::FreeList(uninit_list);
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
bool FdEntity::LoadFull(off_t* size, bool force_load)
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
{
|
|
|
|
int result;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d]", path.c_str(), fd);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 == fd){
|
|
|
|
if(0 != Open()){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(force_load){
|
|
|
|
SetAllDisable();
|
|
|
|
}
|
|
|
|
//
|
|
|
|
// TODO: possibly do background for delay loading
|
|
|
|
//
|
|
|
|
if(0 != (result = Load(0, pagelist.Size()))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("could not download, result(%d)", result);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if(is_modify){
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
is_modify = false;
|
|
|
|
}
|
|
|
|
if(size){
|
|
|
|
*size = pagelist.Size();
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
int FdEntity::RowFlush(const char* tpath, headers_t& meta, bool ow_sse_flg, bool force_sync)
|
|
|
|
{
|
|
|
|
int result;
|
|
|
|
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[tpath=%s][path=%s][fd=%d]", SAFESTRPTR(tpath), path.c_str(), fd);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 == fd){
|
|
|
|
return -EBADF;
|
|
|
|
}
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
|
|
|
|
if(!force_sync && !is_modify){
|
|
|
|
// nothing to update.
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Make decision to do multi upload (or not) based upon file size
|
|
|
|
*
|
|
|
|
* According to the AWS spec:
|
|
|
|
* - 1 to 10,000 parts are allowed
|
|
|
|
* - minimum size of parts is 5MB (expect for the last part)
|
|
|
|
*
|
|
|
|
* For our application, we will define part size to be 10MB (10 * 2^20 Bytes)
|
|
|
|
* maximum file size will be ~64 GB - 2 ** 36
|
|
|
|
*
|
|
|
|
* Initially uploads will be done serially
|
|
|
|
*
|
|
|
|
* If file is > 20MB, then multipart will kick in
|
|
|
|
*/
|
|
|
|
if(pagelist.Size() > MAX_OBJECT_SIZE){ // 64GB - 1
|
|
|
|
// close f ?
|
|
|
|
return -ENOTSUP;
|
|
|
|
}
|
|
|
|
|
|
|
|
// seek to head of file.
|
|
|
|
if(0 != lseek(fd, 0, SEEK_SET)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("lseek error(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(pagelist.Size() >= MULTIPART_LOWLIMIT && !nomultipart){ // 20MB
|
|
|
|
// Additional time is needed for large files
|
|
|
|
time_t backup = 0;
|
|
|
|
if(120 > S3fsCurl::GetReadwriteTimeout()){
|
|
|
|
backup = S3fsCurl::SetReadwriteTimeout(120);
|
|
|
|
}
|
|
|
|
result = S3fsCurl::ParallelMultipartUploadRequest(tpath ? tpath : path.c_str(), meta, fd, ow_sse_flg);
|
|
|
|
if(0 != backup){
|
|
|
|
S3fsCurl::SetReadwriteTimeout(backup);
|
|
|
|
}
|
|
|
|
}else{
|
Fixed Issue 229 and Changes codes
1) Set metadata "Content-Encoding" automatically(Issue 292)
For this issue, s3fs is added new option "ahbe_conf".
New option means the configuration file path, and this file specifies
additional HTTP header by file(object) extension.
Thus you can specify any HTTP header for each object by extension.
* ahbe_conf file format:
-----------
line = [file suffix] HTTP-header [HTTP-header-values]
file suffix = file(object) suffix, if this field is empty,
it means "*"(all object).
HTTP-header = additional HTTP header name
HTTP-header-values = additional HTTP header value
-----------
* Example:
-----------
.gz Content-Encoding gzip
.Z Content-Encoding compress
X-S3FS-MYHTTPHEAD myvalue
-----------
A sample configuration file is uploaded in "test" directory.
If ahbe_conf parameter is specified, s3fs loads it's configuration
and compares extension(suffix) of object(file) when uploading
(PUT/POST) it. If the extension is same, s3fs adds/sends specified
HTTP header and value.
A case of sample configuration file, if a object(it's extension is
".gz") which already has Content-Encoding HTTP header is renamed
to ".txt" extension, s3fs does not set Content-Encoding. Because
".txt" is not match any line in configuration file.
So, s3fs matches the extension by each PUT/POST action.
* Please take care about "Content-Encoding".
This new option allows setting ANY HTTP header by object extension.
For example, you can specify "Content-Encoding" for ".gz"/etc
extension in configuration. But this means that S3 always returns
"Content-Encoding: gzip" when a client requests with other
"Accept-Encoding:" header. It SHOULD NOT be good.
Please see RFC 2616.
2) Changes about allow_other/uid/gid option for mount point
I reviewed about mount point permission and allow_other/uid/gid
options, and found bugs about these.
s3fs is fixed bugs and changed to the following specifications.
* s3fs only allows uid(gid) options as 0(root), when the effective
user is zero(root).
* A mount point(directory) must have a permission to allow
accessing by effective user/group.
* If allow_other option is specified, the mount point permission
is set 0777(all users allow all access).
In another case, the mount point is set 0700(only allows
effective user).
* When uid/gid option is specified, the mount point owner/group
is set uid/gid option value.
If uid/gid is not set, it is set effective user/group id.
This changes maybe fixes some issue(321, 338).
3) Changes a logic about (Issue 229)
The chmod command returns -EIO when changing the mount point.
It is correct, s3fs can not changed owner/group/mtime for the
mount point, but s3fs sends a request for changing the bucket.
This revision does not send the request, and returns EIO as
soon as possible.
git-svn-id: http://s3fs.googlecode.com/svn/trunk@465 df820570-a93a-0410-bd06-b72b767a4274
2013-08-16 19:24:01 +00:00
|
|
|
S3fsCurl s3fscurl(true);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
result = s3fscurl.PutRequest(tpath ? tpath : path.c_str(), meta, fd, ow_sse_flg);
|
|
|
|
}
|
|
|
|
|
|
|
|
// seek to head of file.
|
|
|
|
if(0 == result && 0 != lseek(fd, 0, SEEK_SET)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("lseek error(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(0 == result){
|
|
|
|
is_modify = false;
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
ssize_t FdEntity::Read(char* bytes, off_t start, size_t size, bool force_load)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
int result;
|
|
|
|
ssize_t rsize;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
2013-08-19 06:29:24 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][offset=%jd][size=%zu]", path.c_str(), fd, (intmax_t)start, size);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 == fd){
|
|
|
|
return -EBADF;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(force_load){
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-11-17 08:50:41 +00:00
|
|
|
pagelist.SetInit(start, static_cast<off_t>(size), false);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
// Loading
|
|
|
|
if(0 != (result = Load(start, size))){
|
2013-08-19 06:29:24 +00:00
|
|
|
DPRN("could not download. start(%jd), size(%zu), errno(%d)", (intmax_t)start, size, result);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -EIO;
|
|
|
|
}
|
|
|
|
// Reading
|
|
|
|
{
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(-1 == (rsize = pread(fd, bytes, size, start))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("pread failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return rsize;
|
|
|
|
}
|
|
|
|
|
|
|
|
ssize_t FdEntity::Write(const char* bytes, off_t start, size_t size)
|
|
|
|
{
|
|
|
|
int result;
|
|
|
|
ssize_t wsize;
|
|
|
|
|
2013-08-19 06:29:24 +00:00
|
|
|
FPRNINFO("[path=%s][fd=%d][offset=%jd][size=%zu]", path.c_str(), fd, (intmax_t)start, size);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(-1 == fd){
|
|
|
|
return -EBADF;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Load unitialized area which starts from 0 to (start + size) before writing.
|
|
|
|
if(0 != (result = Load(0, start))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to load uninitialized area before writing(errno=%d)", result);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return static_cast<ssize_t>(result);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Writing
|
|
|
|
{
|
|
|
|
AutoLock auto_lock(&fdent_lock);
|
|
|
|
|
|
|
|
if(-1 == (wsize = pwrite(fd, bytes, size, start))){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("pwrite failed. errno(%d)", errno);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
if(!is_modify){
|
|
|
|
is_modify = true;
|
|
|
|
}
|
|
|
|
if(0 < wsize){
|
2013-11-17 08:50:41 +00:00
|
|
|
pagelist.SetInit(start, static_cast<off_t>(wsize), true);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return wsize;
|
|
|
|
}
|
|
|
|
|
|
|
|
//------------------------------------------------
|
|
|
|
// FdManager class valiable
|
|
|
|
//------------------------------------------------
|
|
|
|
FdManager FdManager::singleton;
|
|
|
|
pthread_mutex_t FdManager::fd_manager_lock;
|
|
|
|
bool FdManager::is_lock_init(false);
|
|
|
|
string FdManager::cache_dir("");
|
|
|
|
size_t FdManager::page_size(FDPAGE_SIZE);
|
|
|
|
|
|
|
|
//------------------------------------------------
|
|
|
|
// FdManager class methods
|
|
|
|
//------------------------------------------------
|
|
|
|
bool FdManager::SetCacheDir(const char* dir)
|
|
|
|
{
|
|
|
|
if(!dir || '\0' == dir[0]){
|
|
|
|
cache_dir = "";
|
|
|
|
}else{
|
|
|
|
cache_dir = dir;
|
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
size_t FdManager::SetPageSize(size_t size)
|
|
|
|
{
|
|
|
|
// If already has entries, this function is failed.
|
|
|
|
if(0 < FdManager::get()->fent.size()){
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
size_t old = FdManager::page_size;
|
|
|
|
FdManager::page_size = size;
|
|
|
|
return old;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool FdManager::DeleteCacheDirectory(void)
|
|
|
|
{
|
|
|
|
if(0 == FdManager::cache_dir.size()){
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
string cache_dir;
|
|
|
|
if(!FdManager::MakeCachePath(NULL, cache_dir, false)){
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
return delete_files_in_dir(cache_dir.c_str(), true);
|
|
|
|
}
|
|
|
|
|
|
|
|
int FdManager::DeleteCacheFile(const char* path)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s]", SAFESTRPTR(path));
|
2013-06-21 06:07:22 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(!path){
|
|
|
|
return -EIO;
|
|
|
|
}
|
|
|
|
if(0 == FdManager::cache_dir.size()){
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
string cache_path = "";
|
|
|
|
if(!FdManager::MakeCachePath(path, cache_path, false)){
|
|
|
|
return 0;
|
|
|
|
}
|
2013-07-29 08:20:19 +00:00
|
|
|
int result = 0;
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
if(0 != unlink(cache_path.c_str())){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("failed to delete file(%s): errno=%d", path, errno);
|
2013-07-29 08:20:19 +00:00
|
|
|
result = -errno;
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
2013-07-29 08:20:19 +00:00
|
|
|
if(!CacheFileStat::DeleteCacheFileStat(path)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNINFO("failed to delete stat file(%s): errno=%d", path, errno);
|
2013-07-29 08:20:19 +00:00
|
|
|
if(0 != errno){
|
|
|
|
result = -errno;
|
|
|
|
}else{
|
|
|
|
result = -EIO;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return result;
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
bool FdManager::MakeCachePath(const char* path, string& cache_path, bool is_create_dir)
|
|
|
|
{
|
|
|
|
if(0 == FdManager::cache_dir.size()){
|
|
|
|
cache_path = "";
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
string resolved_path(FdManager::cache_dir + "/" + bucket);
|
|
|
|
if(is_create_dir){
|
|
|
|
mkdirp(resolved_path + mydirname(path), 0777);
|
|
|
|
}
|
|
|
|
if(!path || '\0' == path[0]){
|
|
|
|
cache_path = resolved_path;
|
|
|
|
}else{
|
|
|
|
cache_path = resolved_path + SAFESTRPTR(path);
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
//------------------------------------------------
|
|
|
|
// FdManager methods
|
|
|
|
//------------------------------------------------
|
|
|
|
FdManager::FdManager()
|
|
|
|
{
|
|
|
|
if(this == FdManager::get()){
|
|
|
|
try{
|
|
|
|
pthread_mutex_init(&FdManager::fd_manager_lock, NULL);
|
|
|
|
FdManager::is_lock_init = true;
|
|
|
|
}catch(exception& e){
|
|
|
|
FdManager::is_lock_init = false;
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNCRIT("failed to init mutex");
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
|
|
|
}else{
|
|
|
|
assert(false);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
FdManager::~FdManager()
|
|
|
|
{
|
|
|
|
if(this == FdManager::get()){
|
|
|
|
for(fdent_map_t::iterator iter = fent.begin(); fent.end() != iter; iter++){
|
|
|
|
FdEntity* ent = (*iter).second;
|
|
|
|
delete ent;
|
|
|
|
}
|
|
|
|
fent.clear();
|
|
|
|
|
|
|
|
if(FdManager::is_lock_init){
|
|
|
|
try{
|
|
|
|
pthread_mutex_destroy(&FdManager::fd_manager_lock);
|
|
|
|
}catch(exception& e){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRNCRIT("failed to init mutex");
|
2013-06-21 06:07:22 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
FdManager::is_lock_init = false;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}else{
|
|
|
|
assert(false);
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
FdEntity* FdManager::GetFdEntity(const char* path)
|
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[path=%s]", SAFESTRPTR(path));
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(!path || '\0' == path[0]){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
AutoLock auto_lock(&FdManager::fd_manager_lock);
|
|
|
|
|
|
|
|
fdent_map_t::iterator iter = fent.find(string(path));
|
|
|
|
if(fent.end() == iter){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
return (*iter).second;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
FdEntity* FdManager::Open(const char* path, off_t size, time_t time, bool force_tmpfile, bool is_create)
|
2013-05-28 05:54:09 +00:00
|
|
|
{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
FdEntity* ent;
|
|
|
|
|
2013-11-17 08:50:41 +00:00
|
|
|
FPRNINFO("[path=%s][size=%jd][time=%jd]", SAFESTRPTR(path), (intmax_t)size, (intmax_t)time);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
if(!path || '\0' == path[0]){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
AutoLock auto_lock(&FdManager::fd_manager_lock);
|
|
|
|
|
|
|
|
fdent_map_t::iterator iter = fent.find(string(path));
|
|
|
|
if(fent.end() != iter){
|
|
|
|
// found
|
|
|
|
ent = (*iter).second;
|
2013-05-28 05:54:09 +00:00
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
}else if(is_create){
|
|
|
|
// not found
|
|
|
|
string cache_path = "";
|
|
|
|
if(!force_tmpfile && !FdManager::MakeCachePath(path, cache_path, true)){
|
2013-08-10 15:29:39 +00:00
|
|
|
DPRN("failed to make cache path for object(%s).", path);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return NULL;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// make new obj
|
|
|
|
ent = new FdEntity(path, cache_path.c_str());
|
|
|
|
fent[string(path)] = ent;
|
|
|
|
|
2013-05-28 05:54:09 +00:00
|
|
|
}else{
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
return NULL;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
|
|
|
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
// open
|
|
|
|
if(-1 == ent->Open(size, time)){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
return ent;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool FdManager::Close(FdEntity* ent)
|
|
|
|
{
|
2013-08-10 15:29:39 +00:00
|
|
|
FPRNINFO("[ent->file=%s][ent->fd=%d]", ent ? ent->GetPath() : "", ent ? ent->GetFd() : -1);
|
Changes codes for performance(part 3)
* Summay
This revision includes big change about temporary file and local cache file.
By this big change, s3fs works with good performance when s3fs opens/
closes/syncs/reads object.
I made a big change about the handling about temporary file and local cache
file to do this implementation.
* Detail
1) About temporary file(local file)
s3fs uses a temporary file on local file system when s3fs does download/
upload/open/seek object on S3.
After this revision, s3fs calls ftruncate() function when s3fs makes the
temporary file.
In this way s3fs can set a file size of precisely length without downloading.
(Notice - ftruncate function is for XSI-compliant systems, so that possibly
you have a problem on non-XSI-compliant systems.)
By this change, s3fs can download a part of a object by requesting with
"Range" http header. It seems like downloading by each block unit.
The default block(part) size is 50MB, it is caused the result which is default
parallel requests count(5) by default multipart upload size(10MB).
If you need to change this block size, you can change by new option
"fd_page_size". This option can take from 1MB(1024 * 1024) to any bytes.
So that, you have to take care about that fdcache.cpp(and fdcache.h) were
changed a lot.
2) About local cache
Local cache files which are in directory specified by "use_cache" option do
not have always all of object data.
This cause is that s3fs uses ftruncate function and reads(writes) each block
unit of a temporary file.
s3fs manages each block unit's status which are "downloaded area" or "not".
For this status, s3fs makes new temporary file in cache directory which is
specified by "use_cache" option. This status files is in a directory which is
named "<use_cache sirectory>/.<bucket_name>/".
When s3fs opens this status file, s3fs locks this file for exclusive control by
calling flock function. You need to take care about this, the status files can
not be laid on network drive(like NFS).
This revision changes about file open mode, s3fs always opens a local cache
file and each status file with writable mode.
Last, this revision adds new option "del_cache", this option means that s3fs
deletes all local cache file when s3fs starts and exits.
3) Uploading
When s3fs writes data to file descriptor through FUSE request, old s3fs
revision downloads all of the object. But new revision does not download all,
it downloads only small percial area(some block units) including writing data
area.
And when s3fs closes or flushes the file descriptor, s3fs downloads other area
which is not downloaded from server. After that, s3fs uploads all of data.
Already r456 revision has parallel upload function, then this revision with
r456 and r457 are very big change for performance.
4) Downloading
By changing a temporary file and a local cache file, when s3fs downloads a
object, it downloads only the required range(some block units).
And s3fs downloads units by parallel GET request, it is same as a case of
uploading. (Maximum parallel request count and each download size are
specified same parameters for uploading.)
In the new revision, when s3fs opens file, s3fs returns file descriptor soon.
Because s3fs only opens(makes) the file descriptor with no downloading
data. And when s3fs reads a data, s3fs downloads only some block unit
including specified area.
This result is good for performance.
5) Changes option name
The option "parallel_upload" which added at r456 is changed to new option
name as "parallel_count". This reason is this option value is not only used by
uploading object, but a uploading object also uses this option. (For a while,
you can use old option name "parallel_upload" for compatibility.)
git-svn-id: http://s3fs.googlecode.com/svn/trunk@458 df820570-a93a-0410-bd06-b72b767a4274
2013-07-23 16:01:48 +00:00
|
|
|
|
|
|
|
AutoLock auto_lock(&FdManager::fd_manager_lock);
|
|
|
|
|
|
|
|
for(fdent_map_t::iterator iter = fent.begin(); iter != fent.end(); iter++){
|
|
|
|
if((*iter).second == ent){
|
|
|
|
ent->Close();
|
|
|
|
if(!ent->IsOpen()){
|
|
|
|
delete (*iter).second;
|
|
|
|
fent.erase(iter);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return false;
|
2013-05-28 05:54:09 +00:00
|
|
|
}
|
|
|
|
|