Successful LRU cache lookups modify the cache's recency list, so
RWMutex.RLock isn't enough protection.
Secondarily, multiple concurrent lookups with the same key should not
create separate rate limiters, so release the lock only when presence
of the key in the cache has been ascertained.
Co-authored-by: greatroar <@>
We set the STRESTART environment when starting the inner process after
the first time, but this didn't persist when restarting the monitor
process. Now it does.
- In the few places where we wrap errors, use the new Go 1.13 "%w"
construction instead of %s or %v.
- Where we create errors with constant strings, consistently use
errors.New and not fmt.Errorf.
- Remove capitalization from errors in the few places where we had that.
I was working on indirecting version vectors, and that resulted in some
refactoring and improving the existing block indirection stuff. We may
or may not end up doing the version vector indirection, but I think
these changes are reasonable anyhow and will simplify the diff
significantly if we do go there. The main points are:
- A bunch of renaming to make the indirection and GC not about "blocks"
but about "indirection".
- Adding a cutoff so that we don't actually indirect for small block
lists. This gets us better performance when handling small files as it
cuts out the indirection for quite small loss in space efficiency.
- Being paranoid and always recalculating the hash on put. This costs
some CPU, but the consequences if a buggy or malicious implementation
silently substituted the block list by lying about the hash would be bad.
Also retain the interval over restarts by storing last GC time in the
database. This to make sure that GC eventually happens even if the
interval is configured to a long time (say, a month).
* lib/db: Deduplicate block lists in database (fixes#5898)
This moves the block list in the database out from being just a field on
the FileInfo to being an object of its own. When putting a FileInfo we
marshal the block list separately and store it keyed by the sha256 of
the marshalled block list. When getting, if we are not doing a
"truncated" get, we do an extra read and unmarshal for the block list.
Old block lists are cleared out by a periodic GC sweep. The alternative
would be to use refcounting, but:
- There is a larger risk of getting that wrong and either dropping a
block list in error or keeping them around forever.
- It's tricky with our current database, as we don't have dirty reads.
This means that if we update two FileInfos with identical block lists in
the same transaction we can't just do read/modify/write for the ref
counters as we wouldn't see our own first update. See above about
tracking this and risks about getting it wrong.
GC uses a bloom filter for keys to avoid heavy RAM usage. GC can't run
concurrently with FileInfo updates so there is a new lock around those
operation at the lowlevel.
The end result is a much more compact database, especially for setups
with many peers where files get duplicated many times.
This is per-key-class stats for a large database I'm currently working
with, under the current schema:
```
0x00: 9138161 items, 870876 KB keys + 7397482 KB data, 95 B + 809 B avg, 1637651 B max
0x01: 185656 items, 10388 KB keys + 1790909 KB data, 55 B + 9646 B avg, 924525 B max
0x02: 916890 items, 84795 KB keys + 3667 KB data, 92 B + 4 B avg, 192 B max
0x03: 384 items, 27 KB keys + 5 KB data, 72 B + 15 B avg, 87 B max
0x04: 1109 items, 17 KB keys + 17 KB data, 15 B + 15 B avg, 69 B max
0x06: 383 items, 3 KB keys + 0 KB data, 9 B + 2 B avg, 18 B max
0x07: 510 items, 4 KB keys + 12 KB data, 9 B + 24 B avg, 41 B max
0x08: 1349 items, 12 KB keys + 10 KB data, 9 B + 8 B avg, 17 B max
0x09: 194 items, 0 KB keys + 123 KB data, 5 B + 634 B avg, 11484 B max
0x0a: 3 items, 0 KB keys + 0 KB data, 14 B + 7 B avg, 30 B max
0x0b: 181836 items, 2363 KB keys + 10694 KB data, 13 B + 58 B avg, 173 B max
Total 10426475 items, 968490 KB keys + 9202925 KB data.
```
Note 7.4 GB of data in class 00, total size 9.2 GB. After running the
migration we get this instead:
```
0x00: 9138161 items, 870876 KB keys + 2611392 KB data, 95 B + 285 B avg, 4788 B max
0x01: 185656 items, 10388 KB keys + 1790909 KB data, 55 B + 9646 B avg, 924525 B max
0x02: 916890 items, 84795 KB keys + 3667 KB data, 92 B + 4 B avg, 192 B max
0x03: 384 items, 27 KB keys + 5 KB data, 72 B + 15 B avg, 87 B max
0x04: 1109 items, 17 KB keys + 17 KB data, 15 B + 15 B avg, 69 B max
0x06: 383 items, 3 KB keys + 0 KB data, 9 B + 2 B avg, 18 B max
0x07: 510 items, 4 KB keys + 12 KB data, 9 B + 24 B avg, 41 B max
0x09: 194 items, 0 KB keys + 123 KB data, 5 B + 634 B avg, 11484 B max
0x0a: 3 items, 0 KB keys + 0 KB data, 14 B + 17 B avg, 51 B max
0x0b: 181836 items, 2363 KB keys + 10694 KB data, 13 B + 58 B avg, 173 B max
0x0d: 44282 items, 1461 KB keys + 61081 KB data, 33 B + 1379 B avg, 1637399 B max
Total 10469408 items, 969939 KB keys + 4477905 KB data.
```
Class 00 is now down to 2.6 GB, with just 61 MB added in class 0d.
There will be some additional reads in some cases which theoretically
hurts performance, but this will be more than compensated for by smaller
writes and better compaction.
On my own home setup which just has three devices and a handful of
folders the difference is smaller in absolute numbers of course, but
still less than half the old size:
```
0x00: 297122 items, 20894 KB keys + 306860 KB data, 70 B + 1032 B avg, 103237 B max
0x01: 115299 items, 7738 KB keys + 17542 KB data, 67 B + 152 B avg, 419 B max
0x02: 1430537 items, 121223 KB keys + 5722 KB data, 84 B + 4 B avg, 253 B max
...
Total 1947412 items, 151268 KB keys + 337485 KB data.
```
to:
```
0x00: 297122 items, 20894 KB keys + 37038 KB data, 70 B + 124 B avg, 520 B max
0x01: 115299 items, 7738 KB keys + 17542 KB data, 67 B + 152 B avg, 419 B max
0x02: 1430537 items, 121223 KB keys + 5722 KB data, 84 B + 4 B avg, 253 B max
...
0x0d: 18041 items, 595 KB keys + 71964 KB data, 33 B + 3988 B avg, 101109 B max
Total 1965447 items, 151863 KB keys + 139628 KB data.
```
* wip
* wip
* wip
* wip
This PR does two things, because one lead to the other:
- Move the leveldb specific stuff into a small "backend" package that
defines a backend interface and the leveldb implementation. This allows,
potentially, in the future, switching the db implementation so another
KV store should we wish to do so.
- Add proper error handling all along the way. The db and backend
packages are now errcheck clean. However, I drew the line at modifying
the FileSet API in order to keep this manageable and not continue
refactoring all of the rest of Syncthing. As such, the FileSet methods
still panic on database errors, except for the "database is closed"
error which is instead handled by silently returning as quickly as
possible, with the assumption that we're anyway "on the way out".
Since we've taken upon ourselves to create a log file by default on
Windows, this adds proper management of that log file. There are two new
options:
-log-max-old-files="3" Number of old files to keep (zero to keep only current).
-log-max-size="10485760" Maximum size of any file (zero to disable log rotation).
The default values result in four files (syncthing.log, synchting.0.log,
..., syncthing.3.log) each up to 10 MiB in size. To not use log rotation
at all, the user can say --log-max-size=0.
This is apparently an old benchmarking tool. I'd forgotten about it.
Since 67b8ef1f3e the build script tries to
build all binaries explicitly by default, and this fails on Windows as
this tool doesn't build on Windows.
Kill it with fire.
This makes sure addresses are sorted when coming in from the API. The
database merge operation still checks for correct ordering (which is
quick) and sorts if it isn't correct (legacy database record or
replication peer), but then does a copy first.
Tested with -race in production...
This adds a certificate lifetime parameter to our certificate generation
and hard codes it to twenty years in some uninteresting places. In the
main binary there are a couple of constants but it results in twenty
years for the device certificate and 820 days for the HTTPS one. 820 is
less than the 825 maximum Apple allows nowadays.
This also means we must be prepared for certificates to expire, so I add
some handling for that and generate a new certificate when needed. For
self signed certificates we regenerate a month ahead of time. For other
certificates we leave well enough alone.
The relay and discosrv didn't use the new lib/build package, now they
do. Conversely the lib/build package wasn't aware there might be other
users and hard coded the program name - now it's set by the build
script
This is the result of:
- Changing build.go to take the protobuf version from the modules
instead of hardcoded
- `go get github.com/gogo/protobuf@v1.3.0` to upgrade
- `go run build.go proto` to regenerate our code
This introduces a better set of defaults for large databases. I've
experimentally determined that it results in much better throughput in a
couple of scenarios with large databases, but I can't give any
guarantees the values are always optimal. They're probably no worse than
the defaults though.
This is a tiny tool to grab the GitHub releases info and generate a
more concise version of it. The conciseness comes from two aspects:
- We select only the latest stable and pre. There is no need to offer
upgrades to versions that are older than the latest. (There might be, in
the future, when we hit 2.0. We can revisit this at that time.)
- We use our structs to deserialize and reserialize the data. This means
we remove all attributes that we don't understand and hence don't
require.
All in all the new response is about 10% the size of the previous one and
avoids the issue where we only serve a bunch of release candidates and
no stable.
This changes the on disk format for new raw reports to be gzip
compressed. Also adds the ability to serve these reports in plain text,
to insulate web browsers from the change (previously we just served the
raw reports from disk using Caddy).
* add skeleton for lib/syncthing
* copy syncthingMain to lib/syncthing (verbatim)
* Remove code to deduplicate copies of syncthingMain
* fix simple build errors
* move stuff from main to syncthing with minimal mod
* merge runtime options
* actually use syncthing.App
* pass io.writer to lib/syncthing for auditing
* get rid of env stuff in lib/syncthing
* add .Error() and comments
* review: Remove fs interactions from lib
* and go 1.13 happened
* utility functions
Per the sync/atomic bug note:
> On ARM, x86-32, and 32-bit MIPS, it is the caller's
> responsibility to arrange for 64-bit alignment of 64-bit words
> accessed atomically. The first word in a variable or in an
> allocated struct, array, or slice can be relied upon to be
> 64-bit aligned.
All atomic accesses of 64-bit variables in syncthing code base are
currently ok (i.e they are all 64-bit aligned).
Generally, the bug is triggered because of incorrect alignement
of struct fields. Free variables (declared in a function) are
guaranteed to be 64-bit aligned by the Go compiler.
To ensure the code remains correct upon further addition/removal
of fields, which would change the currently correct alignment, I
added the following comment where required:
// atomic, must remain 64-bit aligned
See https://golang.org/pkg/sync/atomic/#pkg-note-BUG.
Use a global raven.Client because they allocate an http.Client for each,
with a separate CA bundle and infinite connection idle time. Infinite
connection idle time means that if the client is never used again it
will always keep the connection around, not verifying whether it's
closed server side or not. This leaks about a megabyte of memory for
each client every created.
client.Close() doesn't help with this because the http.Client is still
around, retained by its own goroutines.
The thing with the map is just to retain the API on sendReport, even
though there will in practice only ever be one DSN per process
instance...
* lib/ur: Implement crash (panic) reporting (fixes#959)
This implements a simple crash reporting method. It piggybacks on the
panic log files created by the monitor process, picking these up and
uploading them from the usage reporting routine.
A new config value points to the crash receiver base URL, which defaults
to "https://crash.syncthing.net/newcrash" (following the pattern of
"https://data.syncthing.net/newdata" for usage reports, but allowing us
to separate the service as required).
* cmd/syncthing, lib/gui: Separate gui into own package (ref #4085)
* fix tests
* Don't use main as interface name (make old go happy)
* gui->api
* don't leak state via locations and use in-tree config
* let api (un-)subscribe to config
* interface naming and exporting
* lib/ur
* fix tests and lib/foldersummary
* shorter URVersion and ur debug fix
* review
* model.JsonCompletion(FolderCompletion) -> FolderCompletion.Map()
* rename debug facility https -> api
* folder summaries in model
* disassociate unrelated constants
* fix merge fail
* missing id assignement
* cleanup Fatal in lib/config/config.go
* cleanup Fatal in lib/config/folderconfiguration.go
* cleanup Fatal in lib/model/model.go
* cleanup Fatal in cmd/syncthing/monitor.go
* cleanup Fatal in cmd/syncthing/main.go
* cleanup Fatal in lib/api
* remove Fatal methods from logger
* lowercase in errors.Wrap
* one less channel
I'm working through linter complaints, these are some fixes. Broad
categories:
1) Ignore errors where we can ignore errors: add "_ = ..." construct.
you can argue that this is annoying noise, but apart from silencing the
linter it *does* serve the purpose of highlighting that an error is
being ignored. I think this is OK, because the linter highlighted some
error cases I wasn't aware of (starting CPU profiles, for example).
2) Untyped constants where we though we had set the type.
3) A real bug where we ineffectually assigned to a shadowed err.
4) Some dead code removed.
There'll be more of these, because not all packages are fixed, but the
diff was already large enough.
This adds booleans to the /system/version response to advice the GUI
whether the running version is a candidate release or not. (We could
parse it from the version string, but why duplicate the logic.)
Additionally the settings dialog locks down the upgrade and usage
reporting options on candidate releases. This matches the current
behavior, it just makes it obvious what actually *can* be chosen.
* go mod init; rm -rf vendor
* tweak proto files and generation
* go mod vendor
* clean up build.go
* protobuf literals in tests
* downgrade gogo/protobuf
This changes the TLS and certificate handling in a few ways:
- We always use TLS 1.2, both for sync connections (as previously) and
the GUI/REST/discovery stuff. This is a tightening of the requirements
on the GUI. AS far as I can tell from caniusethis.com every browser from
2013 and forward supports TLS 1.2, so I think we should be fine.
- We always greate ECDSA certificates. Previously we'd create
ECDSA-with-RSA certificates for sync connections and pure RSA
certificates for the web stuff. The new default is more modern and the
same everywhere. These certificates are OK in TLS 1.2.
- We use the Go CPU detection stuff to choose the cipher suites to use,
indirectly. The TLS package uses CPU capabilities probing to select
either AES-GCM (fast if we have AES-NI) or ChaCha20 (faster if we
don't). These CPU detection things aren't exported though, so the tlsutil
package now does a quick TLS handshake with itself as part of init().
If the chosen cipher suite was AES-GCM we prioritize that, otherwise we
prefer ChaCha20. Some might call this ugly. I think it's awesome.
This adds a thin type that holds the state associated with the
leveldb.DB, leaving the huge Instance type more or less stateless. Also
moves some keying stuff into the DB package so that other packages need
not know the keying specifics.
(This does not, yet, fix the cmd/stindex program, in order to keep the
diff size down. Hence the keying constants are still exported.)
* lib/model, cmd/syncthing: Wait for folder restarts to complete (fixes#5233)
This is the somewhat ugly - but on the other hand clear - fix for what
is really a somewhat thorny issue. To avoid zombie folder runners a new
mutex is introduced that protects the RestartFolder operation. I hate
adding more mutexes but the alternatives I can think of are worse.
The other part of it is that the POST /rest/system/config operation now
waits for the config commit to complete. The point of this is that until
the commit has completed we should not accept another config commit. If
we did, we could end up with two separate RestartFolders queued in the
background. While they are both correct, and will run without
interfering with each other, we can't guarantee the order in which they
will run. Thus it could happen that the newer config got committed
first, and the older config commited after that, leaving us with the
wrong config running.
* test
* wip
* hax
* hax
* unflake test
* per folder mutexes
* paranoia
* race
The previous "Bad Request" was really confusing as it implies it's
somethign wrong with the request, which there isn't - the problem is
that server configuration forbids the request.
Adds a receive only folder type that does not send changes, and where the user can optionally revert local changes. Also changes some of the icons to make the three folder types distinguishable.
We have the invalid bit to indicate that a file isn't good. That's enough for remote devices. For ourselves, it would be good to know sometimes why the file isn't good - because it's an unsupported type, because it matches an ignore pattern, or because we detected the data is bad and we need to rescan it.
Or, and this is the main future reason for the PR, because it's a change detected on a receive only device. We will want something like the invalid flag for those changes, but marking them as invalid today means the scanner will rehash them. Hence something more fine grained is required.
This introduces a LocalFlags fields to the FileInfo where we can stash things that we care about locally. For example,
FlagLocalUnsupported = 1 << 0 // The kind is unsupported, e.g. symlinks on Windows
FlagLocalIgnored = 1 << 1 // Matches local ignore patterns
FlagLocalMustRescan = 1 << 2 // Doesn't match content on disk, must be rechecked fully
The LocalFlags fields isn't sent over the wire; instead the Invalid attribute is calculated based on the flags at index sending time. It's on the FileInfo anyway because that's what we serialize to database etc.
The actual Invalid flag should after this just be considered when building the global state and figuring out availability for remote devices. It is not used for local file index entries.
When dropping delta index IDs due to upgrade, only drop our local one.
Previously, when dropping all of them, we would trigger a full send in
both directions on first connect after upgrade. Then the other side
would upgrade, doing the same thing. Net effect is full index data gets
sent twice in both directions.
With this change we just drop our local ID, meaning we will send our
full index on first connect after upgrade. When the other side upgrades,
they will do the same. This is a bit less cruel.
The current 500 "test failed" looks and sounds like a problem in the
relay pool server, while it actually indicates a problem on the
announcing side. Instead use 400 "connection test failed" to indicate
that the request was bad and what was the test.
Since #4340 pulls aren't happening every 10s anymore and may be delayed up to 1h.
This means that no folder error event reaches the web UI for a long time, thus no
failed items will show up for a long time. Now errors are populated when the
web UI is opened.
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4650
LGTM: AudriusButkevicius
This is a new revision of the discovery server. Relevant changes and
non-changes:
- Protocol towards clients is unchanged.
- Recommended large scale design is still to be deployed nehind nginx (I
tested, and it's still a lot faster at terminating TLS).
- Database backend is leveldb again, only. It scales enough, is easy to
setup, and we don't need any backend to take care of.
- Server supports replication. This is a simple TCP channel - protect it
with a firewall when deploying over the internet. (We deploy this within
the same datacenter, and with firewall.) Any incoming client announces
are sent over the replication channel(s) to other peer discosrvs.
Incoming replication changes are applied to the database as if they came
from clients, but without the TLS/certificate overhead.
- Metrics are exposed using the prometheus library, when enabled.
- The database values and replication protocol is protobuf, because JSON
was quite CPU intensive when I tried that and benchmarked it.
- The "Retry-After" value for failed lookups gets slowly increased from
a default of 120 seconds, by 5 seconds for each failed lookup,
independently by each discosrv. This lowers the query load over time for
clients that are never seen. The Retry-After maxes out at 3600 after a
couple of weeks of this increase. The number of failed lookups is
stored in the database, now and then (avoiding making each lookup a
database put).
All in all this means clients can be pointed towards a cluster using
just multiple A / AAAA records to gain both load sharing and redundancy
(if one is down, clients will talk to the remaining ones).
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4648
This keeps the data we need about sequence numbers and object counts
persistently in the database. The sizeTracker is expanded into a
metadataTracker than handled multiple folders, and the Counts struct is
made protobuf serializable. It gains a Sequence field to assist in
tracking that as well, and a collection of Counts become a CountsSet
(for serialization purposes).
The initial database scan is also a consistency check of the global
entries. This shouldn't strictly be necessary. Nonetheless I added a
created timestamp to the metadata and set a variable to compare against
that. When the time since the metadata creation is old enough, we drop
the metadata and rebuild from scratch like we used to, while also
consistency checking.
A new environment variable STCHECKDBEVERY can override this interval,
and for example be set to zero to force the check immediately.
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4547
LGTM: imsodin
So STDEADLOCK seems to do the same thing as STDEADLOCKTIMEOUT, except in
the other package. Consolidate?
STDEADLOCKTHRESHOLD is actually called STLOCKTHRESHOLD, correct the help
text.
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4598
Two issues since the filesystem migration;
- filepath.Base() doesn't do what the code expected when the path ends
in a slash, and we make sure the path ends in a slash.
- the return should be fully qualified, not just the last component.
With this change it works as before, and passes the new test for it.
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4591
LGTM: imsodin
This should address issue as described in https://forum.syncthing.net/t/stun-nig-party-with-paused-devices/10942/13
Essentially the model and the connection service goes out of sync in terms of thinking if we are connected or not.
Resort to model as being the ultimate source of truth.
I can't immediately pin down how this happens, yet some ideas.
ConfigSaved happens in separate routine, so it's possbile that we have some sort of device removed yet connection comes in parallel kind of thing.
However, in this case the connection exists in the model, and does not exist in the connection service and the only way for the connection to be removed
in the connection service is device removal from the config.
Given the subject, this might also be related to the device being paused.
Also, adds more info to the logs
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4533
