This changes the arguments in clock functions to be `Duration` and
converts call sites and constants into `Duration`. There are still some
more functions around that should be converted (e.g.
`timeout_deferred`), but we leave that to another PR.
We also changes `.as_secs()` to return a float, as the rounding broke
things subtly. The only reason to keep it (its the same as
`timedelta.total_seconds()`) is for symmetry with `as_millis()`.
Follows on from https://github.com/element-hq/synapse/pull/19223
### Background
As part of Element's plan to support a light form of vhosting (virtual
host) (multiple instances of Synapse in the same Python process), we're
currently diving into the details and implications of running multiple
instances of Synapse in the same Python process.
"Per-tenant logging" tracked internally by
https://github.com/element-hq/synapse-small-hosts/issues/48
### Prior art
Previously, we exposed `server_name` by providing a static logging
`MetadataFilter` that injected the values:
205d9e4fc4/synapse/config/logger.py (L216)
While this can work fine for the normal case of one Synapse instance per
Python process, this configures things globally and isn't compatible
when we try to start multiple Synapse instances because each subsequent
tenant will overwrite the previous tenant.
### What does this PR do?
We remove the `MetadataFilter` and replace it by tracking the
`server_name` in the `LoggingContext` and expose it with our existing
[`LoggingContextFilter`](205d9e4fc4/synapse/logging/context.py (L584-L622))
that we already use to expose information about the `request`.
This means that the `server_name` value follows wherever we log as
expected even when we have multiple Synapse instances running in the
same process.
### A note on logcontext
Anywhere, Synapse mistakenly uses the `sentinel` logcontext to log
something, we won't know which server sent the log. We've been fixing up
`sentinel` logcontext usage as tracked by
https://github.com/element-hq/synapse/issues/18905
Any further `sentinel` logcontext usage we find in the future can be
fixed piecemeal as normal.
d2a966f922/docs/log_contexts.md (L71-L81)
### Testing strategy
1. Adjust your logging config to include `%(server_name)s` in the format
```yaml
formatters:
precise:
format: '%(asctime)s - %(server_name)s - %(name)s - %(lineno)d -
%(levelname)s - %(request)s - %(message)s'
```
1. Start Synapse: `poetry run synapse_homeserver --config-path
homeserver.yaml`
1. Make some requests (`curl
http://localhost:8008/_matrix/client/versions`, etc)
1. Open the homeserver logs and notice the `server_name` in the logs as
expected. `unknown_server_from_sentinel_context` is expected for the
`sentinel` logcontext (things outside of Synapse).
Spawning from
https://github.com/matrix-org/synapse/pull/12588#discussion_r865843321
> It turns out `Deferred.cancel()` is a lot like
`Deferred.callback()`/`errback()` in that it will trash the logging
context:
> it can resume a coroutine, which will restore its own logging context,
then run:
>
> - until it blocks, setting the sentinel context
> - or until it terminates, setting the context it was started with
>
> So we need to wrap it in `with PreserveLoggingContext():`, like we do
with `.callback()`:
>
> ```python
> with PreserveLoggingContext():
> self.render_deferred.cancel()
> ```
>
> *-- @squahtx,
https://github.com/matrix-org/synapse/pull/12588#discussion_r865843321*
Regressed in
https://github.com/element-hq/synapse/pull/18900#discussion_r2331554278
(see conversation there for more context)
### How is this a regression?
> To give this an update with more hindsight; this logic *was* redundant
with the early return and it is safe to remove this complexity
✅
>
> It seems like this actually has to do with completed vs incomplete
deferreds...
>
> To explain how things previously worked *without* the early-return
shortcut:
>
> With the normal case of **incomplete awaitable**, we store the
`calling_context` and the `f` function is called and runs until it
yields to the reactor. Because `f` follows the logcontext rules, it sets
the `sentinel` logcontext. Then in `run_in_background(...)`, we restore
the `calling_context`, store the current `ctx` (which is `sentinel`) and
return. When the deferred completes, we restore `ctx` (which is
`sentinel`) before yielding to the reactor again (all good
✅)
>
> With the other case where we see a **completed awaitable**, we store
the `calling_context` and the `f` function is called and runs to
completion (no logcontext change). *This is where the shortcut would
kick in but I'm going to continue explaining as if we commented out the
shortcut.* -- Then in `run_in_background(...)`, we restore the
`calling_context`, store the current `ctx` (which is same as the
`calling_context`). Because the deferred is already completed, our extra
callback is called immediately and we restore `ctx` (which is same as
the `calling_context`). Since we never yield to the reactor, the
`calling_context` is perfect as that's what we want again (all good
✅)
>
> ---
>
> But this also means that our early-return shortcut is no longer just
an optimization and is *necessary* to act correctly in the **completed
awaitable** case as we want to return with the `calling_context` and not
reset to the `sentinel` context. I've updated the comment in
https://github.com/element-hq/synapse/pull/18964 to explain the
necessity as it's currently just described as an optimization.
>
> But because we made the same change to
`run_coroutine_in_background(...)` which didn't have the same
early-return shortcut, we regressed the correct behavior ❌ . This is
being fixed in https://github.com/element-hq/synapse/pull/18964
>
>
> *-- @MadLittleMods,
https://github.com/element-hq/synapse/pull/18900#discussion_r2373582917*
### How did we find this problem?
Spawning from @wrjlewis
[seeing](https://matrix.to/#/!SGNQGPGUwtcPBUotTL:matrix.org/$h3TxxPVlqC6BTL07dbrsz6PmaUoZxLiXnSTEY-QYDtA?via=jki.re&via=matrix.org&via=element.io)
`Starting metrics collection 'typing.get_new_events' from sentinel
context: metrics will be lost` in the logs:
<details>
<summary>More logs</summary>
```
synapse.http.request_metrics - 222 - ERROR - sentinel - Trying to stop RequestMetrics in the sentinel context.
2025-09-23 14:43:19,712 - synapse.util.metrics - 212 - WARNING - sentinel - Starting metrics collection 'typing.get_new_events' from sentinel context: metrics will be lost
2025-09-23 14:43:19,713 - synapse.rest.client.sync - 851 - INFO - sentinel - Client has disconnected; not serializing response.
2025-09-23 14:43:19,713 - synapse.http.server - 825 - WARNING - sentinel - Not sending response to request <XForwardedForRequest at 0x7f23e8111ed0 method='POST' uri='/_matrix/client/unstable/org.matrix.simplified_msc3575/sync?pos=281963%2Fs929324_147053_10_2652457_147960_2013_25554_4709564_0_164_2&timeout=30000' clientproto='HTTP/1.1' site='8008'>, already dis
connected.
2025-09-23 14:43:19,713 - synapse.access.http.8008 - 515 - INFO - sentinel - 92.40.194.87 - 8008 - {@me:wi11.co.uk} Processed request: 30.005sec/-8.041sec (0.001sec, 0.000sec) (0.000sec/0.002sec/2) 0B 200! "POST /_matrix/client/unstable/org.matrix.simplified_msc3575/
```
</details>
From the logs there, we can see things relating to
`typing.get_new_events` and
`/_matrix/client/unstable/org.matrix.simplified_msc3575/sync` which led
me to trying out Sliding Sync with the typing extension enabled and
allowed me to reproduce the problem locally. Sliding Sync is a unique
scenario as it's the only place we use `gather_optional_coroutines(...)`
-> `run_coroutine_in_background(...)` (introduced in
https://github.com/element-hq/synapse/pull/17884) to exhibit this
behavior.
### Testing strategy
1. Configure Synapse to enable
[MSC4186](https://github.com/matrix-org/matrix-spec-proposals/pull/4186):
Simplified Sliding Sync which is actually under
[MSC3575](https://github.com/matrix-org/matrix-spec-proposals/pull/3575)
```yaml
experimental_features:
msc3575_enabled: true
```
1. Start synapse: `poetry run synapse_homeserver --config-path
homeserver.yaml`
1. Make a Sliding Sync request with one of the extensions enabled
```http
POST
http://localhost:8008/_matrix/client/unstable/org.matrix.simplified_msc3575/sync
{
"lists": {},
"room_subscriptions": {
"!FlgJYGQKAIvAscfBhq:my.synapse.linux.server": {
"required_state": [],
"timeline_limit": 1
}
},
"extensions": {
"typing": {
"enabled": true
}
}
}
```
1. Open your homeserver logs and notice warnings about `Starting ...
from sentinel context: metrics will be lost`
Part of https://github.com/element-hq/synapse/issues/18905
Lints for ensuring we use `Clock.call_later` instead of
`reactor.callLater`, etc are coming in
https://github.com/element-hq/synapse/pull/18944
### Testing strategy
1. Configure Synapse to log at the `DEBUG` level
1. Start Synapse: `poetry run synapse_homeserver --config-path
homeserver.yaml`
1. Wait 10 seconds for the [database profiling
loop](9cc4001778/synapse/storage/database.py (L711))
to execute
1. Notice the logcontext being used for the `Total database time` log
line
Before (`sentinel`):
```
2025-09-10 16:36:58,651 - synapse.storage.TIME - 707 - DEBUG - sentinel - Total database time: 0.646% {room_forgetter_stream_pos(2): 0.131%, reap_monthly_active_users(1): 0.083%, get_device_change_last_converted_pos(1): 0.078%}
```
After (`looping_call`):
```
2025-09-10 16:36:58,651 - synapse.storage.TIME - 707 - DEBUG - looping_call - Total database time: 0.646% {room_forgetter_stream_pos(2): 0.131%, reap_monthly_active_users(1): 0.083%, get_device_change_last_converted_pos(1): 0.078%}
```
Introduce `Clock.call_when_running(...)` to wrap startup code in a
logcontext, ensuring we can identify which server generated the logs.
Background:
> Ideally, nothing from the Synapse homeserver would be logged against the `sentinel`
> logcontext as we want to know which server the logs came from. In practice, this is not
> always the case yet especially outside of request handling.
>
> Global things outside of Synapse (e.g. Twisted reactor code) should run in the
> `sentinel` logcontext. It's only when it calls into application code that a logcontext
> gets activated. This means the reactor should be started in the `sentinel` logcontext,
> and any time an awaitable yields control back to the reactor, it should reset the
> logcontext to be the `sentinel` logcontext. This is important to avoid leaking the
> current logcontext to the reactor (which would then get picked up and associated with
> the next thing the reactor does).
>
> *-- `docs/log_contexts.md`
Also adds a lint to prefer `Clock.call_when_running(...)` over
`reactor.callWhenRunning(...)`
Part of https://github.com/element-hq/synapse/issues/18905
During the migration the automated script to update the copyright
headers accidentally got rid of some of the existing copyright lines.
Reinstate them.
* Pull Sentinel out of LoggingContext
... and drop a few unnecessary references to it
* Factor out LoggingContext.current_context
move `current_context` and `set_context` out to top-level functions.
Mostly this means that I can more easily trace what's actually referring to
LoggingContext, but I think it's generally neater.
* move copy-to-parent into `stop`
this really just makes `start` and `stop` more symetric. It also means that it
behaves correctly if you manually `set_log_context` rather than using the
context manager.
* Replace `LoggingContext.alive` with `finished`
Turn `alive` into `finished` and make it a bit better defined.
when processing incoming transactions, it can be hard to see what's going on,
because we process a bunch of stuff in parallel, and because we may end up
recursively working our way through a chain of three or four events.
This commit creates a way to use logcontexts to add the relevant event ids to
the log lines.
So, it turns out that if you have a first `Deferred` `D1`, you can add a
callback which returns another `Deferred` `D2`, and `D2` must then complete
before any further callbacks on `D1` will execute (and later callbacks on `D1`
get the *result* of `D2` rather than `D2` itself).
So, `D1` might have `called=True` (as in, it has started running its
callbacks), but any new callbacks added to `D1` won't get run until `D2`
completes - so if you `yield D1` in an `inlineCallbacks` function, your `yield`
will 'block'.
In conclusion: some of our assumptions in `logcontext` were invalid. We need to
make sure that we don't optimise out the logcontext juggling when this
situation happens. Fortunately, it is easy to detect by checking `D1.paused`.
It turns out that the only thing we use the __dict__ of LoggingContext for is
`request`, and given we create lots of LoggingContexts and then copy them every
time we do a db transaction or log line, using the __dict__ seems a bit
redundant. Let's try to optimise things by making the request attribute
explicit.
