The old version was taken from an MPL-licensed library. This is a
cleanroom implementation. The code is shorter and it's now explicit that
only Linux ACLs are supported.
The test uses `WithTimeout` to create a context that cancels the List
operation after a given delay. Several backends internally use a derived
child context created using WithCancel.
The cancellation of a context first closes the done channel of the
context (here: the `WithTimeout` context) and _afterwards_ propagates
the cancellation to child contexts (here: the `WithCancel` context).
Therefor if the List implementation uses a child context, then it may
take a moment until that context is also cancelled. Thus give the
context cancellation a moment to propagate.
A stale lock may be refreshed if it continues to exist until after a
replacement lock has been created. This ensures that a repository was
not unlocked in the meantime.
When transferring a repository from S3 to, for example, a local disk
then all empty folders will be missing.
When saving files, the missing intermediate folders are created
automatically. Therefore, missing directories can be ignored by the
`List()` operation.
Linux allows the use of non-`user.` extended attributes on symlinks. One
of the main users of this functionality is SELinux's `security.selinux`
xattr for storing a path's label. By storing symlink xattrs, restic is
now suitable for backing up the root filesystem on Linux distributions
that use SELinux.
This commit adds support for symlink xattrs when backing up data,
restoring data, and mounting snapshots via a fuse mount. All calls to
the xattr library have been updated to the use `L` variants of the
various functions, which always operate on the path given, without
following symlinks.
Fixes: #4375
Signed-off-by: Andrew Gunnerson <accounts+github@chiller3.com>
Conceptually the backend configuration should be validated when creating
or opening the backend, but not when filling in information from
environment variables into the configuration.
This unified construction removes most backend-specific code from
global.go. The backend registry will also enable integration tests to
use custom backends if necessary.
The ETA restic displays was based on a rate computed across the entire
backup operation. Often restic can progress at uneven rates. In the worst
case, restic progresses over most of the backup at a very high rate and
then finds new data to back up. The displayed ETA is then unrealistic and
never adapts.
Restic now estimates the transfer rate based on a sliding window, with the
goal of adapting to observed changes in rate. To avoid wild changes in the
estimate, several heuristics are used to keep the sliding window wide
enough to be relatively stable.
In order to change the backend initialization in `global.go` to be able
to generically call cfg.ApplyEnvironment() for supported backends, the
`interface{}` returned by `ParseConfig` must contain a pointer to the
configuration.
An alternative would be to use reflection to convert the type from
`interface{}(Config)` to `interface{}(*Config)` (from value to pointer
type). However, this would just complicate the type mess further.
Iterating through the indexmap according to the bucket order has the
problem that all indexEntries are accessed in random order which is
rather cache inefficient.
As we already keep a list of all allocated blocks, just iterate through
it. This allows iterating through a batch of indexEntries without random
memory accesses. In addition, the packID will likely remain similar
across multiple blobs as all blobs of a pack file are added as a single
batch.
This data structure reduces the wasted memory to O(sqrt(n)). The
top-layer of the hashed array tree (HAT) also has a size of O(sqrt(n)),
which makes it cache efficient. The top-layer should be small enough to
easily fit into the CPU cache and thus only adds little overhead
compared to directly accessing an index entry via a pointer.
The indexEntry objects are now allocated in a separate array. References
to an indexEntry are now stored as array indices. This has the benefit
of allowing the garbage collector to ignore the indexEntry objects as
these do not contain pointers and are part of a single large allocation.
New and its helpers used to create the cache directories several times
over. They now only do so once. The added test ensures that the cache is
produced in a consistent state when parts are deleted.
Use the logging methods from testing.TB to make use of tb.Helper(). This
allows the tests to log the filename and line number in which the test
helper was called. Previously the test helper was logged which is rarely
useful.
This function casts its argument to int32 before passing it to the
system call, so that big-endian CPUs read the lower rather than the
upper 32 bits of the pid.
This also gets rid of the last import of "unsafe" in the Unix build.
I changed syscall to x/sys/unix while I was at it, to remove one more
import line. The constants and types there are aliases for their syscall
counterparts.
As the `Fatal` error type only includes a string, it becomes impossible
to inspect the contained error. This is for a example a problem for the
fuse implementation, which must be able to detect context.Canceled
errors.
Co-authored-by: greatroar <61184462+greatroar@users.noreply.github.com>
For hardlinked files, only the first instance of that file increases the
amount of bytes to restore. All later instances only increase the file
count but not the restore size.
restic must be able to refresh lock files in time. However, large
uploads over slow connections can cause the lock refresh to be stuck
behind the large uploads and thus time out.
The test had a 4% chance of not modified the data read from the
repository, in which case the test would fail. Change the data
manipulation to just modified each read operation.
This adds support for caching already rewritten trees, handling of load
errors and disabling the check that the serialization doesn't lead to
data loss.
The more generic RewriteNode callback replaces the SelectByName and
PrintExclude functions. The main part of this change is a preparation to
allow using the TreeRewriter for the `repair snapshots` command.
The builtin mechanism to capture a stacktrace in Go is to send a SIGQUIT
to the running process. However, this mechanism is not avaiable on
Windows. Thus, tweak the SIGINT handler to dump a stacktrace if the
environment variable `RESTIC_DEBUG_STACKTRACE_SIGINT` is set.
The SemaphoreBackend now uniformly enforces the limit of concurrent
backend operations. In addition, it unifies the parameter validation.
The List() methods no longer uses a semaphore. Restic already never runs
multiple list operations in parallel.
By managing the semaphore in a wrapper backend, the sections that hold a
semaphore token grow slightly. However, the main bottleneck is IO, so
this shouldn't make much of a difference.
The key insight that enables the SemaphoreBackend is that all of the
complex semaphore handling in `openReader()` still happens within the
original call to `Load()`. Thus, getting and releasing the semaphore
tokens can be refactored to happen directly in `Load()`. This eliminates
the need for wrapping the reader in `openReader()` to release the token.
The snapshot filtering internally converts relative paths to absolute
ones to ensure that the parent snapshots selection works for backups of
relative paths.
x/text/width.LookupRune has to re-encode its argument as UTF-8,
while LookupString operates on the UTF-8 directly.
The uint casts get rid of a bounds check.
Benchmark results, with b.ResetTimer introduced first:
name old time/op new time/op delta
TruncateASCII-8 69.7ns ± 1% 55.2ns ± 1% -20.90% (p=0.000 n=20+18)
TruncateUnicode-8 350ns ± 1% 171ns ± 1% -51.05% (p=0.000 n=20+19)
Since 0.15 (#4020), inodes are generated as hashes of names, xor'd with
the parent inode. That means that the inode of a/b/b is
h(a/b/b) = h(a) ^ h(b) ^ h(b) = h(a).
I.e., the grandchild has the same inode as the grandparent. GNU find
trips over this because it thinks it has encountered a loop in the
filesystem, and fails to search a/b/b. This happens more generally when
the same name occurs an even number of times.
Fix this by multiplying the parent by a large prime, so the combining
operation is not longer symmetric in its arguments. This is what the FNV
hash does, which we used prior to 0.15. The hash is now
h(a/b/b) = h(b) ^ p*(h(b) ^ p*h(a))
Note that we already ensure that h(x) is never zero.
Collisions can still occur, but they should be much less likely to occur
within a single path.
Fixes #4253.
Tests in cmd_forget_test.go need the same convenience function
that was implemented in snapshot_policy_test.go.
Function parseDuration(...) was moved to testing.go and renamed to
ParseDurationOrPanic(...).
This turns snapshotFilterOptions from cmd into a restic.SnapshotFilter
type and makes restic.FindFilteredSnapshot and FindFilteredSnapshots
methods on that type. This fixes #4211 by ensuring that hosts and paths
are named struct fields instead of unnamed function arguments in long
lists of such.
Timestamp limits are also included in the new type. To avoid too much
pointer handling, the convention is that time zero means no limit.
That's January 1st, year 1, 00:00 UTC, which is so unlikely a date that
we can sacrifice it for simpler code.
This method had a buffer argument, but that was nil at all call sites.
That's removed, and instead LoadUnpacked now reuses whatever it
allocates inside its retry loop.
With debug logging enabled this method would take a lock and then
format the lock as a string. Since PR #4022 landed the string
formatting method has also taken the lock, so this deadlocks.
Instead just record the lock ID, as is done elsewhere.
The code always assumed that the upgrade happens in place. Thus writing
the upgrade to a separate file fails, when trying to remove the file
stored at that location.
Added missing call to scanFinished=true.
This was causing the percent and eta to never get
printed for backup progress even after the scan was finished.
The retry backend does not return the original error, if its execution
is interrupted by canceling the context. Thus, we have to manually
ensure that the invalid data error gets returned.
Additionally, use the retry backend for some of the repository tests, as
this is the configuration which will be used by restic.
The retry printed the filename twice:
```
Load(<lock/04804cba82>, 0, 0) returned error, retrying after 720.254544ms: load(<lock/04804cba82>): invalid data returned
```
now the warning has changed to
```
Load(<lock/04804cba82>, 0, 0) returned error, retrying after 720.254544ms: invalid data returned
```
The StdioWrapper is not used at all by the ProgressPrinters. It is
called a bit earlier than previously. However, as the password prompt
directly accessed stdin/stdout this doesn't cause problems.
Mostly changed the ones that repeat the name of a system call, which is
already contained in os.PathError.Op. internal/fs.Reader had to be
changed to actually return such errors.
TestRepository and its variants always returned no-op cleanup functions.
If they ever do need to do cleanup, using testing.T.Cleanup is easier
than passing these functions around.
We now check for space that is not reserved for the root user on the
remote, and the check is no longer in a defer block because it wouldn't
fire. Some change in the surrounding code may have led the deferred
function to capture the wrong err variable.
Fixes #3336.
IDs.Less can be rewritten as
string(list[i][:]) < string(list[j][:])
Note that this does not copy the ID's.
The Uniq method was no longer used.
The String method has been reimplemented without first copying into a
separate slice of a custom type.
The Test method was only used in exactly one place, namely when trying
to create a new repository it was used to check whether a config file
already exists.
Use a combination of Stat() and IsNotExist() instead.
Since #3940 the rclone backend returns the commands exit code if it
fails to start. The list of expected errors was missing the "file
already closed"-error which can occur if the http test request first
learns about the closed pipe to rclone before noticing the canceled
context.
Go internally makes sure that a file descriptor is unusable once it was
closed, thus this cannot have unintended side effects (like accidentally
reading from the wrong file due to a reused file descriptor).
The ioutil functions are deprecated since Go 1.17 and only wrap another
library function. Thus directly call the underlying function.
This commit only mechanically replaces the function calls.
Without comma-ok, the runtime inserts the same check with a similar
enough panic message:
interface conversion: interface {} is nil, not *syscall.Stat_t
The new genericized LRU cache no longer needs to have the IDs separately
allocated:
name old time/op new time/op delta
Add-8 494ns ± 2% 388ns ± 2% -21.46% (p=0.000 n=10+9)
name old alloc/op new alloc/op delta
Add-8 176B ± 0% 152B ± 0% -13.64% (p=0.000 n=10+10)
name old allocs/op new allocs/op delta
Add-8 5.00 ± 0% 3.00 ± 0% -40.00% (p=0.000 n=10+10)
Hard links to the same file now get the same inode within the FUSE
mount. Also, inode generation is faster and, more importantly, no longer
allocates.
Benchmarked on Linux/amd64. Old means the benchmark with
sink = fs.GenerateDynamicInode(1, sub.node.Name)
instead of calling inodeFromNode. Results:
name old time/op new time/op delta
Inode/no_hard_links-8 137ns ± 4% 34ns ± 1% -75.20% (p=0.000 n=10+10)
Inode/hard_link-8 33.6ns ± 1% 9.5ns ± 0% -71.82% (p=0.000 n=9+8)
name old alloc/op new alloc/op delta
Inode/no_hard_links-8 48.0B ± 0% 0.0B -100.00% (p=0.000 n=10+10)
Inode/hard_link-8 0.00B 0.00B ~ (all equal)
name old allocs/op new allocs/op delta
Inode/no_hard_links-8 1.00 ± 0% 0.00 -100.00% (p=0.000 n=10+10)
Inode/hard_link-8 0.00 0.00 ~ (all equal)
In principle, the JSON format of Tree objects is extensible without
requiring a format change. In order to not loose information just play
it safe and reject rewriting trees for which we could loose data.
The lock test creates a lock and checks that it is not stale. However,
it is possible that the lock is refreshed concurrently, which updates
the lock timestamp. Checking the timestamp in `Stale()` without
synchronization results in a data race. Thus add a lock to prevent
concurrent accesses.
The lock test creates a lock and checks that it is not stale. This also
tests whether the corresponding process still exists. However, it is
possible that the lock is refreshed concurrently, which updates the lock
timestamp. Calling `processExists()` with a value receiver, however,
creates an unsynchronized copy of this field. Thus call the method using
a pointer receiver.
In some rare cases files could be created which contain null IDs (all
zero) in their content list. This was caused by a race condition between
growing the `Content` slice and inserting the blob IDs into it. In some
cases the blob ID was written to the old slice, which a short time
afterwards was replaced with a larger copy, that did not yet contain the
blob ID.
Automatically fall back to hiding files if not authorized to permanently
delete files. This allows using restic with an append-only application
key with B2. Thus, an attacker cannot directly delete backups with the
API key used by restic.
To use this feature create an application key without the deleteFiles
capability. It is recommended to restrict the key to just one bucket.
For example using the b2 command line tool:
b2 create-key --bucket <bucketName> <keyName> listBuckets,readFiles,writeFiles,listFiles
Suggested-by: Daniel Gröber <dxld@darkboxed.org>
We previously checked whether the set of snapshots might have changed
based only on their number, which fails when as many snapshots are
forgotten as are added. Check for the SHA-256 of their id's instead.
The status bar got stuck once the first error was reported, the scanner
completed or some file was backed up. Either case sets a flag that the
scanner has started.
This flag is used to hide the progress bar until the flag is set. Due to
an inverted condition, the opposite happened and the status stopped
refreshing once the flag was set.
In addition, the scannerStarted flag was not set when the scanner just
reported progress information.
As the FileSaver is asynchronously waiting for all blobs of a file to be
stored, the number of active files is higher than the number of files
from which restic is reading concurrently. Thus to not confuse users,
only display files in the status from which restic is currently reading.
After reading and chunking all data in a file, the FutureFile still has
to wait until the FutureBlobs are completed. This was done synchronously
which results in blocking the file saver and prevents the next file from
being read.
By replacing the FutureBlob with a callback, it becomes possible to
complete the FutureFile asynchronously.
We always need both values, except in a test, so we don't need to lock
twice and risk scheduling in between.
Also, removed the resetting in Done. This copied a mutex, which isn't
allowed. Static analyzers tend to trip over that.
The channel-based algorithm had grown quite complicated. This is easier
to reason about and likely to be more performant with very many
CompleteBlob calls.
As long as only a small fraction of the data in a repository is
rewritten, the keepBlobs set will be rather small after cleaning it up.
As golang maps do not shrink their memory usage, just copy the contents
over to a new map. However, only copy the map if the cleanup removed at
least half the entries.
The set covers necessary, existing and duplicate blobs. This removes the
duplicate sets used to track whether all necessary blobs also exist.
This reduces the memory usage of prune by about 20-30%.
The RetryBackend tests depend on the mock backend. When the Backend
interface is eventually split from the restic package, this will lead to
a dependency cycle between backend and backend/mock. Thus split the
RetryBackend into a separate package to avoid this problem.
Archiver.Save queries the current time multiple times. This commit
removes one of these calls as they showed up while profiling a backup of
a nearly unchanged dataset containing 3 million files.
The string form was presumably useful before the introduction of
layouts, but right now it just makes call sequences and garbage
collection more expensive (the latter because every string contains
a pointer to be scanned).
if x { return true } return false => return x
fmt.Sprintf("%v", x) => fmt.Sprint(x) or x.String()
The fmt.Sprintf idiom is still used in the SecretString tests, where it
serves security hardening.
ID.UnmarshalJSON accepted non-JSON input with ' as the string delimiter.
Also, the error message for non-hex input was less informative than it
could be and it performed too many checks.
Changed ParseID to keep the error messages consistent.
FindFilteredSnapshots no longer prints errors during snapshot loading on
stderr, but instead passes the error to the callback to allow the caller
to decide on what to do.
In addition, it moves the logic to handle an explicit snapshot list from
the main package to restic.
The helper function uidGidInt used strconv.ParseInt instead of
ParseUint, so it silently ignored some invalid user/group IDs.
Also, improve the error message. "Invalid UID" is more informative than
having "ParseInt" twice (*strconv.NumError displays the function name).
Finally, the user.User struct can be passed by pointer to get reduce
code size.
This package is no longer needed, since we can use the stdlib's
http.NewRequestWithContext.
backend/rclone already did, but it needed a different error check due to
a difference between net/http and ctxhttp.
Also, store the http.Client by value in the REST backend (changed to a
pointer when ctxhttp was introduced) and use errors.WithStack instead
of errors.Wrap where the message was no longer accurate. Errors from
http.NewRequestWithContext will start with "net/http" or "net/url", so
they're easy to identify.
The backup command failed if a directory contains duplicate entries.
Downgrade the severity of this problem from fatal error to a warning.
This allows users to still create a backup.
SaveTree did not use the TreeSaver but rather managed the tree
collection and upload itself. This prevents using the parallelism
offered by the TreeSaver and duplicates all related code. Using the
TreeSaver can provide some speed-ups as all steps within the backup tree
now rely on FutureNodes. This can be especially relevant for backups
with large amounts of explicitly specified files.
The main difference between SaveTree and SaveDir is, that only the
former can save tree blobs in which nodes have a different name than the
actual file on disk. This is the result of resolving name conflicts
between multiple files with the same name. The filename that must be
used within the snapshot is now passed directly to
restic.NodeFromFileInfo. This ensures that a FutureNode already contains
the correct filename.
According to the documentation of exec.Cmd Wait() must not be called
before completing all reads from the pipe returned by StdErrPipe(). Thus
return a context that is canceled once rclone has exited and use that as
a precondition to calling Wait(). This should ensure that all errors
printed to stderr have been copied first.
When rclone fails during the connection setup this currently often
results in a context canceled error. Replace this error with the exit
code from rclone.
When backing up many small files, the unbuffered channels frequently
cause the FileSaver to block when reporting progress information. Thus,
add buffers to these channels to avoid unnecessary scheduling.
As the status information is purely informational, it doesn't matter
that the status reporting shutdown is somewhat racy and could miss a few
final updates.
The only use cases in the code were in errors.IsFatal, backend/b2,
which needs a workaround, and backend.ParseLayout. The last of these
requires all backends to implement error unwrapping in IsNotExist.
All backends except gs already did that.
Repositories with mixed packs are probably quite rare by now. When
loading data blobs from a mixed pack file, this will no longer trigger
caching that file. However, usually tree blobs are accessed first such
that this shouldn't make much of a difference.
The checker gets a simpler replacement.
While searching for lock file from concurrently running restic
instances, restic ignored unreadable lock files. These can either be
in fact invalid or just be temporarily unreadable. As it is not really
possible to differentiate between both cases, just err on the side of
caution and consider the repository as already locked.
The code retries searching for other locks up to three times to smooth
out temporarily unreadable lock files.
Restic continued e.g. a backup task even when it failed to renew the
lock or failed to do so in time. For example if a backup client enters
standby during the backup this can allow other operations like `prune`
to run in the meantime (after calling `unlock`). After leaving standby
the backup client will continue its backup and upload indexes which
refer pack files that were removed in the meantime.
This commit introduces a goroutine explicitly monitoring for locks that
are not refreshed in time. To simplify the implementation there's now a
separate goroutine to refresh the lock and monitor for timeouts for each
lock. The monitoring goroutine would now cause the backup to fail as the
client has lost it's lock in the meantime.
The lock refresh goroutines are bound to the context used to lock the
repository initially. The context returned by `lockRepo` is also
cancelled when any of the goroutines exits. This ensures that the
context is cancelled whenever for any reason the lock is no longer
refreshed.
Some backends generate additional files for each existing file, e.g.
1234567890abcdef1234567890abcdef1234567890abcdef1234567890abcdef
1234567890abcdef1234567890abcdef1234567890abcdef1234567890abcdef.sha256
For some commands this leads to an "multiple IDs with prefix" error when
trying to reference a snapshot.
Failing to process data requested from the cache usually indicates a
problem with the returned data. Assume that the cache entry is somehow
damaged and retry downloading it once.
Sparse files contain large regions containing only zero bytes. Checking
that a blob only contains zeros is possible with over 100GB/s for modern
x86 CPUs. Calculating sha256 hashes is only possible with 500MB/s (or
2GB/s using hardware acceleration). Thus we can speed up the hash
calculation for all zero blobs (which always have length
chunker.MinSize) by checking for zero bytes and then using the
precomputed hash.
The all zeros check is only performed for blobs with the minimal chunk
size, and thus should add no overhead most of the time. For chunks which
are not all zero but have the minimal chunks size, the overhead will be
below 2% based on the above performance numbers.
This allows reading sparse sections of files as fast as the kernel can
return data to us. On my system using BTRFS this resulted in about
4GB/s.
The restorer can issue multiple calls to WriteAt in parallel. This can
result in unexpected orderings of the Truncate and WriteAt calls and
sometimes too short restored files.
We can either preallocate storage for a file or sparsify it. This
detects a pack file as sparse if it contains an all zero block or
consists of only one block. As the file sparsification is just an
approximation, hide it behind a `--sparse` parameter.
This writes files by using (*os.File).Truncate, which resolves to the
truncate system call on Unix.
Compared to the naive loop,
for _, b := range p {
if b != 0 {
return false
}
}
the optimized allZero is about 10× faster:
name old time/op new time/op delta
AllZero-8 1.09ms ± 1% 0.09ms ± 1% -92.10% (p=0.000 n=10+10)
name old speed new speed delta
AllZero-8 3.84GB/s ± 1% 48.59GB/s ± 1% +1166.51% (p=0.000 n=10+10)
`restic unlock` now only shows `successfully removed locks` if there were locks to be removed.
In addition, it also reports the number of the removed lock files.
Sending data through a channel at very high frequency is extremely
inefficient. Thus use simple callbacks instead of channels.
> name old time/op new time/op delta
> MasterIndexEach-16 6.68s ±24% 0.96s ± 2% -85.64% (p=0.008 n=5+5)
This is especially useful if ssh asks for a password or if closing the
initial connection could return an error due to a problematic server
implementation.
bazil/fuse expects us to return context.Canceled to signal that a
syscall was successfully interrupted. Returning a wrapped version of
that error however causes the fuse library to signal an EIO (input/output
error). Thus unwrap context.Canceled errors before returning them.
rclone can exit early for example when the connection to rclone is
relayed for example via ssh: `-o rclone.program='ssh user@example.org
forced-command'`
For backends which are able to atomically replace files, we just can
overwrite the old copy, if it is necessary to retry an upload. This has
the benefit of issuing one operation less and might be beneficial if a
backend storage, due to bugs or similar, could mix up the order of the
upload and delete calls.
When hard deleting the latest file version on B2, this uncovers earlier
versions. If an upload required retries, multiple version might exist
for a file. Thus to reliably delete a file, we have to remove all
versions of it.
Ignored packs were reported as an empty pack by EachByPack. The most
immediate effect of this is that the progress bar for rebuilding the
index reports processing more packs than actually exist.
Previously the buffer was grown incrementally inside `repo.LoadUnpacked`.
But we can do better as we already know how large the index will be.
Allocate a bit more memory to increase the chance that the buffer can be
reused in the future.
Instead of first checking whether a file is in the repository cache and
then opening it, we just can open the file. This saves one stat call. If
the file is in the cache, everything is fine and otherwise the code
follows its normal fallback path.
sort.Sort is not guaranteed to be stable. Go 1.19 has changed the
sorting algorithm which resulted in changes of the sort order. When
comparing snapshots with identical timestamp but different paths and
tags lists, there is not meaningful order among them. So just keep their
order stable.
Cleanly separate the directory presentation and the snapshot directory
structure. SnapshotsDir now translates the dirStruct into a format
usable by the fuse library and contains only minimal special case rules.
All decisions have moved into SnapshotsDirStructure which now creates a
fully preassembled tree data structure.
For large pack sizes we might be only interested in the first and last
blob of a pack file. Thus stream a pack file in multiple parts if the
gaps between requested blobs grow too large.
Also make the errors a bit less verbose by not prepending the operation,
since pkg/xattr already does that. Old errors looked like
Listxattr: xattr.list /myfiles/.zfs/snapshot: invalid argument
pkg/sftp.Client.MkdirAll(d) does a Stat to determine if d exists and is
a directory, then a recursive call to create the parent, so the calls
for data/?? each take three round trips. Doing a Mkdir first should
eliminate two round trips for 255/256 data directories as well as all
but one of the top-level directories.
Also, we can do all of the calls concurrently. This may reintroduce some
of the Stat calls when multiple goroutines try to create the same
parent, but at the default number of connections, that should not be
much of a problem.
FutureBlob now uses a Take() method as a more memory-efficient way to
retrieve the futures result. In addition, futures are now collected
while saving the file. As only a limited number of blobs can be queued
for uploading, for a large file nearly all FutureBlobs already have
their result ready, such that the FutureBlob object just consumes
memory.
There is no real difference between the FutureTree and FutureFile
structs. However, differentiating both increases the size of the
FutureNode struct.
The FutureNode struct is now only 16 bytes large on 64bit platforms.
That way is has a very low overhead if the corresponding file/directory
was not processed yet.
There is a special case for nodes that were reused from the parent
snapshot, as a go channel seems to have 96 bytes overhead which would
result in a memory usage regression.
