Flattened the copier code more. Also removing and moving some
parameters/return values to simplify things. Generally rely less on
return values, e.g. by handling errors right away and using `state` to
do the right thing (e.g. abort on failure).
Supposed to be a refactor without any behaviour changes, except for
fixing a tiny regression on folder order: We used to try copying from
the same folder first, but lost that property at some point (also sent a
PR fixing only that, I'd merge that first making this refactor only).
Where `folderFilesystems` and `folders` is built, there's a comment
spelling out the purpose: To have the same folder first, as that's the
most likely to get hits. Plus a copy is possibly more efficient than
from another folder, e.g. if that's on a different filesystem. We lost
that behaviour during some unrelated change.
(Also sneaking in a comment fix on yesterdays change.)
This is a draft because I haven't adjusted all the tests yet, I'd like
to get feedback on the change overall first, before spending time on
that.
In my opinion the main win of this change is in it's lower complexity
resp. fewer moving parts. It should also be faster as it only does one
query instead of two, but I have no idea if that's practically
relevant.
This also mirrors the v1 DB, where a block map key had the name
appended. Not that this is an argument for the change, it was mostly
reassuring me that I might not be missing something key here
conceptually (I might still be of course, please tell me :) ).
And the change isn't mainly intrinsically motivated, instead it came
up while fixing a bug in the copier. And the nested nature of that code
makes the fix harder, and "un-nesting" it required me to understand
what's happening. This change fell out of that.
After opening the database, we performed some checks, such as whether
the migration had already been successfully completed. If so, the
function returned immediately, and the database was not closed.
---------
Co-authored-by: Jakob Borg <jakob@kastelo.net>
### Purpose
Fix https://github.com/syncthing/syncthing/issues/9336
The `emitLoginAttempt` function now checks for the presence of an
`X-Forwarded-For` header. The IP from this header is only used if the
connecting host is either on loopback or on the same LAN.
In the case of a host pretending to be a proxy, we'd still have both IPs
in the logs, which should make this much less critical from a security
standpoint.
### Testing
1. directly via localhost
2. via proxy an localhost
#### Logs
```
[3JPXJ] 2025/04/11 15:00:40 INFO: Wrong credentials supplied during API authorization from 127.0.0.1
[3JPXJ] 2025/04/11 15:03:04 INFO: Wrong credentials supplied during API authorization from 192.168.178.5 proxied by 127.0.0.1
```
#### Event API
```
{
"id": 23,
"globalID": 23,
"time": "2025-04-11T15:00:40.578577402+02:00",
"type": "LoginAttempt",
"data": {
"remoteAddress": "127.0.0.1",
"success": false,
"username": "sdfsd"
}
},
{
"id": 24,
"globalID": 24,
"time": "2025-04-11T15:03:04.423403976+02:00",
"type": "LoginAttempt",
"data": {
"proxy": "127.0.0.1",
"remoteAddress": "192.168.178.5",
"success": false,
"username": "sdfsd"
}
}
```
### Documentation
https://github.com/syncthing/docs/pull/907
---------
Co-authored-by: Jakob Borg <jakob@kastelo.net>
This cleans up the option to allow old TLS 1.2 sync connections. The
flag existed for compatibility with old Syncthing versions that don't
support TLS 1.3, which is approximately Syncthing 1.2.2 (September 2019)
and older. ("Approximately" because it depends on the Go version it's
built with and that's when we switched to building with Go 1.13.)
Ref #10062 because it reminded me this exists.
* main:
build: push artifacts to Azure (#10044)
fix(syncthing): use separate lock file instead of locking the certificate (fixes#10053) (#10054)
fix(syncthing): use separate lock file instead of locking the certificate (fixes#10053) (#10054)
* main:
fix(gui): fix previous commit
fix(gui): mark unseen disconnected devices as inactive (#10048)
fix(strings): differentiate setup(n) and set(v) up (#10024)
chore(fs): changes to allow Filesystem to be implemented externally (#10040)
chore(config): resolve primary STUN servers via SRV record (fixes#10029) (#10031)
build: push artifacts to Azure (#10044)
chore(gui, man, authors): update docs, translations, and contributors
### Purpose
The `fs.Filesystem` interface contains two parts that cannot be
implemented externally because they are private:
* `filesystemWrapperType`: this PR changes `unwrapFilesystem` to
downcast to a specific concrete type
* `underlying`: this PR simply moves it to an unexported interface
### Testing
Regular tests pass.
This adds a simple delay to the process for starting the pull, by
default one second. In practice this means we're likely to wait for
initial index transfer, or multiple messages sent as part of a larger
change. This is better because we're more likely to have the whole
change for the purpose of handling renames etc, and also it's more
efficient to do one larger puller iteration instead of multiple while
also processing changes.
It does however introduce a certain amount of delay into the sync
process, so it can be tuned down or turned off entirely.
This changes the database structure to use one database per folder, with
a small main database to coordinate. Reverts the prior change to buffer
all files in memory when pulling, meaning there is now a phase where the
WAL file will grow significantly, at least for initial sync of folders
with many directories.
---------
Co-authored-by: bt90 <btom1990@googlemail.com>
* main:
fix(config): properly apply defaults when reading folder configuration (#10034)
chore(model): add metric for total number of conflicts (#10037)
build: replace underscore in Debian version (#10032)
This reduces the number of file entries we carry in the database,
sometimes significantly. The downside is that if a file is deleted while
a device is offline, and that device comes back more than the cutoff
interval (six months) later, those files will get resurrected at some
point.
Similarly to #10009, we will remove some discontinued STUN servers,
except instead of being the official primary server, it's some
unofficial secondary STUN servers.
### Testing
Use a STUN client (like [`pystun3`](https://pypi.org/project/pystun3))
to probe that the removed STUN servers are inactive.
### Documentation
syncthing/docs#902
The mechanism for primary STUN servers, is still intact, in case this
gets retried with a different domain.
### Purpose
As seen in [stun.syncthing.net doesn’t resolve
anymore](https://forum.syncthing.net/t/stun-syncthing-net-doesnt-resolve-anymore/24075/2?u=marbens)
on the forums, stun.syncthing.net has been shut down, so I think it's
probably a good idea to remove it.
### Testing
1. Have two or more devices
2. Disable Relaying
3. Have no Internet ports open on either end for incoming connections
trigger STUN)
4. Enable the `stun` debugging facility in the Actions -> Logs ->
Debugging Facilities
5. Verify that it doesn't output something like this within a few
seconds:
```
2025-03-30 05:51:32 Enabled debug data for "stun"
2025-03-30 05:51:47 Starting stun for Stun@udp://[::]:22000
2025-03-30 05:51:47 Running stun for Stun@udp://[::]:22000 via stun.syncthing.net:3478
2025-03-30 05:51:47 Stun@udp://[::]:22000 stun addr resolution on stun.syncthing.net:3478: lookup stun.syncthing.net: no such host
```
---------
Co-authored-by: Jakob Borg <jakob@kastelo.net>
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.
### Purpose
This exposes four methods from `Model` through `Internals`. It allows
apps like Synctrain to obtain information about local/remote need and
sync progress.
### Testing
No testing seems necessary, functions are exported verbatim.
### Screenshots
N/a
### Documentation
Not public API, I am aware this interface may change at any time.
## Authorship
OK.
Co-authored-by: Ross Smith II <ross@smithii.com>
Co-authored-by: Jakob Borg <jakob@kastelo.net>
### Purpose
Path autocompletion wasn't working when using `~` as a shortcut for the
home directory. The issue occurred because the tilde was expanded to
/home/user, which caused the suggestion to no longer match the input
(thus preventing the autocompletion from appearing in the suggestion
list).
To fix this, I replaced the custom `parentAndBase` function, which
handled path splitting in a more complex way, with `filepath.Split` from
the standard `path/filepath` package. This prevents tilde expansion
while keeping the expected behavior for path splitting.
### Testing
The issue has been tested manually on Linux.
### Screenshots
This is extracted from PR #9175. This deduplicates `SetPassword` calls
and makes `postAdjustGui` a single place where PR #9175 can add
another adjustment step for sanitizing changes to WebAuthn
credentials.
This also adds tests to validate that the refactored logic was not
broken.
Simon FreiandGitHub
Marcus B SpencerandGitHub
TheCreeperandGitHub
Sébastien WENSKEandGitHub
Paul DonaldandGitHub
Tommy van der VorstandGitHub
Marcus B SpencerandGitHub
mathias4833andGitHub
Emil LundbergandGitHub