Currently, the number of hashers is always set to 1 on interactive
operating systems, which are defined as Windows, macOS, iOS, and
Android. However, with modern multicore CPUs, it does not make much
sense to limit performance so much.
For example, without this fix, a CPU with 16 cores / 32 threads is
still limited to using just a single thread to hash files per folder,
which may severely affect its performance.
For this reason, instead of using a fixed value, calculate the number
dynamically, so that it equals one-fourth of the total number of CPU
cores. This way, the value of hashes will still end up being just 1 on
a slower 4-thread CPU, but it will be allowed to take larger values when
the number of threads is higher, increasing hashing performance in the
process.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Co-authored-by: Jakob Borg <jakob@kastelo.net>
Currently, the number of hashers, with the exception of some specific
operating systems or when defined manually, equals the number of CPU
cores divided by the overall number of folders, and it does not take
into account the value of MaxFolderConcurrency at all. This leads to
artificial performance limits even when MaxFolderConcurrency is set to
values lower than the number of cores.
For example, let's say that the number of folders is 50 and
MaxFolderConcurrency is set a value of 4 on a 16-core CPU. With the old
calculation, the number of hashers would still end up being just 1 due
to the large number of folders. However, with the new calculation, the
number of hashers in this case will be 4, leading to better hashing
performance per folder.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Co-authored-by: Jakob Borg <jakob@kastelo.net>
This updates our logging framework from legacy freetext strings using
the `log` package to structured log entries using `log/slog`. I have
updated all INFO or higher level entries, but not yet DEBUG (😓)... So,
at a high level:
There is a slight change in log levels, effectively adding a new warning
level:
- DEBUG is still debug (ideally not for users but developers, though
this is something we need to work on)
- INFO is still info, though I've added more data here, effectively
making Syncthing more verbose by default (more on this below)
- WARNING is a new log level that is different from the _old_ WARNING
(more below)
- ERROR is what was WARNING before -- problems that must be dealt with,
and also bubbled as a popup in the GUI.
A new feature is that the logging level can be set per package to
something other than just debug or info, and hence I feel that we can
add a bit more things into INFO while moving some (in fact, most)
current INFO level warnings into WARNING. For example, I think it's
justified to get a log of synced files in INFO and sync failures in
WARNING. These are things that have historically been tricky to debug
properly, and having more information by default will be useful to many,
while still making it possible get close to told level of inscrutability
by setting the log level to WARNING. I'd like to get to a stage where
DEBUG is never necessary to just figure out what's going on, as opposed
to trying to narrow down a likely bug.
Code wise:
- Our logging object, generally known as `l` in each package, is now a
new adapter object that provides the old API on top of the newer one.
(This should go away once all old log entries are migrated.) This is
only for `l.Debugln` and `l.Debugf`.
- There is a new level tracker that keeps the log level for each
package.
- There is a nested setup of handlers, since the structure mandated by
`log/slog` is slightly convoluted (imho). We do this because we need to
do formatting at a "medium" level internally so we can buffer log lines
in text format but with separate timestamp and log level for the API/GUI
to consume.
- The `debug` API call becomes a `loglevels` API call, which can set the
log level to `DEBUG`, `INFO`, `WARNING` or `ERROR` per package. The GUI
is updated to handle this.
- Our custom `sync` package provided some debugging of mutexes quite
strongly integrated into the old logging framework, only turned on when
`STTRACE` was set to certain values at startup, etc. It's been a long
time since this has been useful; I removed it.
- The `STTRACE` env var remains and can be used the same way as before,
while additionally permitting specific log levels to be specified,
`STTRACE=model:WARN,scanner:DEBUG`.
- There is a new command line option `--log-level=INFO` to set the
default log level.
- The command line options `--log-flags` and `--verbose` go away, but
are currently retained as hidden & ignored options since we set them by
default in some of our startup examples and Syncthing would otherwise
fail to start.
Sample format messages:
```
2009-02-13 23:31:30 INF A basic info line (attr1="val with spaces" attr2=2 attr3="val\"quote" a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with grouped values (attr1=val1 foo.attr2=2 foo.bar.attr3=3 a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with grouped values via logger (foo.attr1=val1 foo.attr2=2 a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with nested grouped values via logger (bar.foo.attr1=val1 bar.foo.attr2=2 a=a log.pkg=slogutil)
2009-02-13 23:31:30 WRN A warning entry (a=a log.pkg=slogutil)
2009-02-13 23:31:30 ERR An error (a=a log.pkg=slogutil)
```
---------
Co-authored-by: Ross Smith II <ross@smithii.com>
This makes a couple of backwards compatible changes to the
ClusterConfig:
- Remove the `ignore_permissions` and `ignore_delete` booleans which
we've never read or used for anything
- Remove the `disable_temp_indexes` boolean and option entirely. We did
use this one, and about 1% of users have set the option. The only thing
it does is inhibits sending of periodical DownloadProgress messages
while downloading data, which is a minuscule bandwidth optimisation
given that we're already sending data at the time.
- Change the `read_only` boolean (which indicated send-only folders) to
an enum `FolderType`, where the values zero and one match the existing
usage. Again, we don't actually use this value, but I can see that we
might want to and then it makes more sense for it to be more
comprehensive.
- Change the `paused` boolean to an enum `StopReason`, where zero
indicates not stopped and one indicates paused, exactly the same wire
representation as previously but leaves space for additional stop
reasons (errors etc).
* main:
feat(config): expose folder and device info as metrics (fixes#9519) (#10148)
chore: add issue types to GitHub issue templates
build: remove schedule from PR metadata job
chore(protocol): only allow enc. password changes on cluster config (#10145)
chore(protocol): don't start connection routines a second time (#10146)
In practice we already always call SetPassword and ClusterConfig
together. However it's not just "sensible" to do that, it's required: If
the passwords change, the remote device needs to know about that to
check that the enc. setup is valid/consistent (e.g. tokens match,
folder-type is appropriate, ...).
And with the passwords set later, there's no point in adding them as
part of creating a new connection.
This is a "followup" (if one can call it that 4 years later :) ) to
resp. fix for the following commit:
924b96856f
Co-authored-by: Jakob Borg <jakob@kastelo.net>
* main:
fix(syncthing): ensure both config and data dirs exist at startup (fixes#10126) (#10127)
fix(versioner): fix perms of created folders (fixes#9626) (#10105)
refactor: use slices.Contains to simplify code (#10121)
Switch the database from LevelDB to SQLite, for greater stability and
simpler code.
Co-authored-by: Tommy van der Vorst <tommy@pixelspark.nl>
Co-authored-by: bt90 <btom1990@googlemail.com>
We've had weak/rolling hashing in the code for quite a while. It was a
popular request for a while, based on the belief that rsync does this
and we should too. However, the benefit is quite small; we save on
average about 0.8% of transferred blocks over the population as a whole:
<img width="974" alt="Screenshot 2025-03-28 at 17 09 02"
src="https://github.com/user-attachments/assets/bbe10dea-f85e-4043-9823-7cef1220b4a2"
/>
This would be fine if the cost was comparably low, however the downside
of attempting rolling hash matching is that we (by default) do a
complete file read on the destination in order to look for matches
before we starting pulling blocks for the file. For any larger file this
means a sometimes long, I/O-intensive pause before the file starts
syncing, for usually no benefit.
I propose we simply rip off the bandaid and save the effort.
This was broken in the Protobuf refactor. The reason is that with the
previous library, generated types would have a JSON marshalling method
that would automatically return strings for enums. In the current
library you need to use the jsonpb marshaller for that, but these are
hand crafted structs so we can't do that. The easy solution is to just
use strings directly, since this is an API-only type anyway.
At a high level, this is what I've done and why:
- I'm moving the protobuf generation for the `protocol`, `discovery` and
`db` packages to the modern alternatives, and using `buf` to generate
because it's nice and simple.
- After trying various approaches on how to integrate the new types with
the existing code, I opted for splitting off our own data model types
from the on-the-wire generated types. This means we can have a
`FileInfo` type with nicer ergonomics and lots of methods, while the
protobuf generated type stays clean and close to the wire protocol. It
does mean copying between the two when required, which certainly adds a
small amount of inefficiency. If we want to walk this back in the future
and use the raw generated type throughout, that's possible, this however
makes the refactor smaller (!) as it doesn't change everything about the
type for everyone at the same time.
- I have simply removed in cold blood a significant number of old
database migrations. These depended on previous generations of generated
messages of various kinds and were annoying to support in the new
fashion. The oldest supported database version now is the one from
Syncthing 1.9.0 from Sep 7, 2020.
- I changed config structs to be regular manually defined structs.
For the sake of discussion, some things I tried that turned out not to
work...
### Embedding / wrapping
Embedding the protobuf generated structs in our existing types as a data
container and keeping our methods and stuff:
```
package protocol
type FileInfo struct {
*generated.FileInfo
}
```
This generates a lot of problems because the internal shape of the
generated struct is quite different (different names, different types,
more pointers), because initializing it doesn't work like you'd expect
(i.e., you end up with an embedded nil pointer and a panic), and because
the types of child types don't get wrapped. That is, even if we also
have a similar wrapper around a `Vector`, that's not the type you get
when accessing `someFileInfo.Version`, you get the `*generated.Vector`
that doesn't have methods, etc.
### Aliasing
```
package protocol
type FileInfo = generated.FileInfo
```
Doesn't help because you can't attach methods to it, plus all the above.
### Generating the types into the target package like we do now and
attaching methods
This fails because of the different shape of the generated type (as in
the embedding case above) plus the generated struct already has a bunch
of methods that we can't necessarily override properly (like `String()`
and a bunch of getters).
### Methods to functions
I considered just moving all the methods we attach to functions in a
specific package, so that for example
```
package protocol
func (f FileInfo) Equal(other FileInfo) bool
```
would become
```
package fileinfos
func Equal(a, b *generated.FileInfo) bool
```
and this would mostly work, but becomes quite verbose and cumbersome,
and somewhat limits discoverability (you can't see what methods are
available on the type in auto completions, etc). In the end I did this
in some cases, like in the database layer where a lot of things like
`func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv
*generated.FileVersion)` because they were anyway just internal methods.
Fixes#8247
I came accross this in another context and didn't investigate fully, but
literally ten lines above this code, in another method, we say that
filesets _must_ be created under the lock. It's either one or the other
and I'm taking the safer route here.
---------
Co-authored-by: Simon Frei <freisim93@gmail.com>
### Purpose
As discussed in #9686
Syncthing currently does not check folderstate on remote device before
pulling. If no devices have a valid folderstate (i.e all devices have
the folder paused) it will still attempt to pull. On large folders this
will cause a hanging "Syncing" status.
This checks whether at least one connected device has the file available
and has a valid folderstate.
### Testing
Tested locally on multiple devices.
We're new to Go (all our stuff is Python) so please bear with!
Interested if there may be a better place to slot this in.
Thanks,
Jon
---------
Co-authored-by: Simon Frei <freisim93@gmail.com>
This should prevent the panic that occurred in this test run:
https://github.com/syncthing/syncthing/actions/runs/11095876010/job/30825046810
```
2024-09-29T21:01:53.5425372Z === RUN TestIssue4357
2024-09-29T21:01:53.5505943Z panic: runtime error: integer divide by zero [recovered]
2024-09-29T21:01:53.5512200Z panic: runtime error: integer divide by zero
2024-09-29T21:01:53.5516633Z
2024-09-29T21:01:53.5523018Z goroutine 2655 [running]:
2024-09-29T21:01:53.5524157Z github.com/thejerf/suture/v4.(*Supervisor).runService.func2.2()
2024-09-29T21:01:53.5527176Z /home/runner/go/pkg/mod/github.com/thejerf/suture/v4@v4.0.5/supervisor.go:563 +0xd0
2024-09-29T21:01:53.5530556Z panic({0x1080d20?, 0x1851290?})
2024-09-29T21:01:53.5564723Z /home/runner/go/pkg/mod/golang.org/toolchain@v0.0.1-go1.23.1.linux-amd64/src/runtime/panic.go:785 +0x132
2024-09-29T21:01:53.5566616Z github.com/syncthing/syncthing/lib/model.(*model).numHashers(0xc0006f6180, {0x117dc1a, 0x7})
2024-09-29T21:01:53.5568061Z /home/runner/work/syncthing/syncthing/lib/model/model.go:2581 +0x210
2024-09-29T21:01:53.5569912Z github.com/syncthing/syncthing/lib/model.(*folder).scanSubdirsChangedAndNew(0xc00c38c808, {0x0, 0x0, 0x0}, 0xc0003fc060)
2024-09-29T21:01:53.5571612Z /home/runner/work/syncthing/syncthing/lib/model/folder.go:653 +0x250
2024-09-29T21:01:53.5573010Z github.com/syncthing/syncthing/lib/model.(*folder).scanSubdirs(0xc00c38c808, {0x0, 0x0, 0x0})
2024-09-29T21:01:53.5574447Z /home/runner/work/syncthing/syncthing/lib/model/folder.go:512 +0xd0f
2024-09-29T21:01:53.5576011Z github.com/syncthing/syncthing/lib/model.(*folder).scanTimerFired(0xc00c38c808)
2024-09-29T21:01:53.5577367Z /home/runner/work/syncthing/syncthing/lib/model/folder.go:916 +0x46
2024-09-29T21:01:53.5579010Z github.com/syncthing/syncthing/lib/model.(*folder).Serve(0xc00c38c808, {0x1307650, 0xc0006a0910})
2024-09-29T21:01:53.5580428Z /home/runner/work/syncthing/syncthing/lib/model/folder.go:205 +0xd7e
2024-09-29T21:01:53.5581624Z github.com/thejerf/suture/v4.(*Supervisor).runService.func2()
2024-09-29T21:01:53.5582978Z /home/runner/go/pkg/mod/github.com/thejerf/suture/v4@v4.0.5/supervisor.go:567 +0x249
2024-09-29T21:01:53.5584400Z created by github.com/thejerf/suture/v4.(*Supervisor).runService in goroutine 2651
2024-09-29T21:01:53.5585872Z /home/runner/go/pkg/mod/github.com/thejerf/suture/v4@v4.0.5/supervisor.go:541 +0x32a
2024-09-29T21:01:53.5661413Z FAIL github.com/syncthing/syncthing/lib/model 5.510s
```
### Testing
I have not been able to reproduce the panic throughout a few minutes of
continuously running the test without this fix, but judging by the
traceback it seems to only happen if the test happens to delete the
folder from config at the same time `scanTimerFired` triggers.
### Purpose
This PR contains the set of changes needed to make Syncthing work on iOS
for [my iOS app for
Syncthing](https://github.com/pixelspark/sushitrain).
Most changes originate from [the Mobius Sync
fork](http://github.com/MobiusSync/syncthing/tree/ios). I have removed
the changes from their fork that are not strictly needed for my app
(i.e. their changes to the GUI and command line utilities, for instance)
and squashed it all in a single commit.
In summary, the changes are:
* Resolve non-absolute paths to the 'Documents' folder (basically the
only one an app can/should write user data to by default on iOS)
* Tweaking of build flags/conditions for iOS (i.e. determine which
basicfs_watch, ignoreresult variant to build for iOS)
* Disable upgrade mechanism on iOS
* Make `RequestGlobal` and `PullerProgress` public symbols
* Expose syncthing.app's Model instance (app.M)
* Add no-op stub for SetLowPriority on iOS
I would very much appreciate these changes to be (eventually) merged to
mainline syncthing, as this would allow my iOS app to track the mainline
source code directly and removes the need (for me at least) for
maintaining a separate fork. Perhaps the Mobius folks can also benefit
from this (although as noted this branch does not contain their changes
to e.g. the GUI).
### Testing
This branch has been tested with the iOS app and appears to work fine.
The full set of MobiusSync changes has been used before with success.
### Screenshots
n/a
### Documentation
There should be no visible changes for users due to this set of changes.
---------
Co-authored-by: Simon Pickup <simon@pickupinfinity.com>
### Purpose
Avoid the issue where the folder marker is deleted by overzealous
cleanup tools because it's just a useless, empty directory.
We create a small file containing a an admonishment to not delete the
directory, and some metadata that is just for human consumption at the
moment. (But it would parse as a valid yaml file if we wanted to read
this, at some point.)
This will only apply when _creating_ a folder marker, that is, existing
setups will not gain the file automatically. Obviously, when using a
custom folder marker none of this applies.
Also, slightly adjust the permission bits for the folder marker directory and file on Unixes, making sure the group & write bits are unset.
### Testing
I've created and deleted a few folders and it appears to behave as I
expect.
### Screenshots
```
jb@ok:~/somefolder % ls -la
total 0
drwxr-xr-x 3 jb staff 96 May 1 08:52 ./
drwx------ 12 jb staff 384 May 1 08:52 ../
drwxr-xr-x 3 jb staff 96 May 1 08:52 .stfolder/
jb@ok:~/somefolder % ls -l .stfolder
total 8
-rw-r--r-- 1 jb staff 122 May 1 08:52 syncthing-folder-39a4b0.txt
jb@ok:~/somefolder % cat .stfolder/syncthing-folder-39a4b0.txt
# This directory is a Syncthing folder marker.
# Do not delete.
folderID: xtdca-cudyf
created: 2024-05-01T08:52:49+02:00
```
This is a refactor of the protocol/model interface to take the actual
message as the parameter, instead of the broken-out fields:
```diff
type Model interface {
// An index was received from the peer device
- Index(conn Connection, folder string, files []FileInfo) error
+ Index(conn Connection, idx *Index) error
// An index update was received from the peer device
- IndexUpdate(conn Connection, folder string, files []FileInfo) error
+ IndexUpdate(conn Connection, idxUp *IndexUpdate) error
// A request was made by the peer device
- Request(conn Connection, folder, name string, blockNo, size int32, offset int64, hash []byte, weakHash uint32, fromTemporary bool) (RequestResponse, error)
+ Request(conn Connection, req *Request) (RequestResponse, error)
// A cluster configuration message was received
- ClusterConfig(conn Connection, config ClusterConfig) error
+ ClusterConfig(conn Connection, config *ClusterConfig) error
// The peer device closed the connection or an error occurred
Closed(conn Connection, err error)
// The peer device sent progress updates for the files it is currently downloading
- DownloadProgress(conn Connection, folder string, updates []FileDownloadProgressUpdate) error
+ DownloadProgress(conn Connection, p *DownloadProgress) error
}
```
(and changing the `ClusterConfig` to `*ClusterConfig` for symmetry;
we'll be forced to use all pointers everywhere at some point anyway...)
The reason for this is that I have another thing cooking which is a
small troubleshooting change to check index consistency during transfer.
This required adding a field or two to the index/indexupdate messages,
and plumbing the extra parameters in umpteen changes is almost as big a
diff as this is. I figured let's do it once and avoid having to do that
in the future again...
The rest of the diff falls out of the change above, much of it being in
test code where we run these methods manually...
Cleanup after #9275.
This renames `fmut` -> `mut`, removes the deadlock detector and
associated plumbing, renames some things from `...PRLocked` to
`...RLocked` and similar, and updates comments.
Apart from the removal of the deadlock detection machinery, no
functional code changes... i.e. almost 100% diff noise, have fun
reviewing.
I'm tired of the fmut/pmut shenanigans. This consolidates both under one
lock; I'm not convinced there are any significant performance
differences with this approach since we're literally just protecting map
juggling...
- The locking goes away when we were already under an appropriate fmut
lock.
- Where we had fmut.RLock()+pmut.Lock() it gets upgraded to an
fmut.Lock().
- Otherwise s/pmut/fmut/.
In order to avoid diff noise for an important change I did not do the
following cleanups, which will be filed in a PR after this one, if
accepted:
- Renaming fmut to just mut
- Renaming methods that refer to being "PRLocked" and stuff like that
- Removing the no longer relevant deadlock detector
- Comments referring to pmut and locking sequences...
Before introducing the service map and using it for folder runners, the
entries in folderCfgs and folderRunners for the same key/folder were
removed under a single lock. Stopping the folder happens separately
before that with just the read lock. Now with the service map stopping
the folder and removing it from the map is a single operation. And that
still happens with just a read-lock. However even with a full lock it’s
still problematic: After the folder stopped, the runner isn’t present
anymore while the folder-config still is and sais the folder isn't
paused.
The index handler in turn looks at the folder config that is not paused,
thus assumes the runner has to be present -> nil deref on the runner.
A better solution might be to push most of these fmut maps into the
folder - they anyway are needed in there. Then there's just a single
map/source of info that's necessarily consistent. That's quite a bit of
work though, and probably/likely there will be corner cases there too.
LastSeen for a device was only updated when they connected. This now
updates it when they disconnect, so that we remember the last time we
actually saw them. When asking the API for current stats, currently
connected devices get a last seen value of the current time.
This adds the ability to have multiple concurrent connections to a single device. This is primarily useful when the network has multiple physical links for aggregated bandwidth. A single connection will never see a higher rate than a single link can give, but multiple connections are load-balanced over multiple links.
It is also incidentally useful for older multi-core CPUs, where bandwidth could be limited by the TLS performance of a single CPU core -- using multiple connections achieves concurrency in the required crypto calculations...
Co-authored-by: Simon Frei <freisim93@gmail.com>
Co-authored-by: tomasz1986 <twilczynski@naver.com>
Co-authored-by: bt90 <btom1990@googlemail.com>
Instead of separately tracking the token.
Also changes serviceMap to have a channel version of RemoveAndWait, so
that it's possible to do the removal under a lock but wait outside of
the lock. And changed where we do that in connection close, reversing
the change that happened when I added the serviceMap in 40b3b9ad1.
* lib/versioner: Factor out DefaultPath constant.
Replace several instances where .stversions is named literally to all
use the same definition in the versioner package. Exceptions are the
packages where a cyclic dependency on versioner is impossible, or some
tests which combine the versions base path with other components.
* lib/versioner: Fix comment about trash can in simple versioner.
* lib/versioner: Fix wrong versioning type string in error message.
The error message shows the folder type instead of the versioning
type, although the correct field is used in the comparison.
This fixes various test issues with Go 1.20.
- Most tests rewritten to use fakefs where possible
- Some tests that were already skipped, or dubious (invasive,
unmaintainable, unclear what they even tested) have been removed
- Some actual code rewritten to better support testing in fakefs
Co-authored-by: Eric P <eric@kastelo.net>
This adds a cache to the expensive key generation operations. It's fixes
size LRU/MRU stuff to keep memory usage bounded under absurd conditions.
Also closes#8600.
