We were sending device updates down both the federation stream and
device streams. This mean there was a race if the federation sender
worker processed the federation stream first, as when the sender checked
if there were new device updates the slaved ID generator hadn't been
updated with the new stream IDs and so returned nothing.
This situation is correctly handled by events/receipts/etc by not
sending updates down the federation stream and instead having the
federation sender worker listen on the other streams and poke the
transaction queues as appropriate.
* Port synapse.replication.tcp to async/await
* Newsfile
* Correctly document type of on_<FOO> functions as async
* Don't be overenthusiastic with the asyncing....
ExpiringCache required that `start()` be called before it would actually
start expiring entries. A number of places didn't do that.
This PR removes `start` from ExpiringCache, and automatically starts
backround reaping process on creation instead.
We should explicitly close any db connections we open, because failing to do so
can block other transactions as per
https://github.com/matrix-org/synapse/issues/3682.
Let's also try to factor out some of the boilerplate by having server classes
define their datastore class rather than duplicating the whole of `setup`.
Run the handlers for replication commands as background processes. This should
improve the visibility in our metrics, and reduce the number of "running db
transaction from sentinel context" warnings.
Ideally it means converting the things that fire off deferreds into the night
into things that actually return a Deferred when they are done. I've made a bit
of a stab at this, but it will probably be leaky.
(instead of everywhere that writes a response. Or rather, the subset of places
which write responses where we haven't forgotten it).
This also means that we don't have to have the mysterious version_string
attribute in anything with a request handler.
Unfortunately it does mean that we have to pass the version string wherever we
instantiate a SynapseSite, which has been c&ped 150 times, but that is code
that ought to be cleaned up anyway really.
While I was going through uses of preserve_fn for other PRs, I converted places
which only use the wrapped function once to use run_in_background, to avoid
creating the function object.
There were a bunch of places where we fire off a process to happen in the
background, but don't have any exception handling on it - instead relying on
the unhandled error being logged when the relevent deferred gets
garbage-collected.
This is unsatisfactory for a number of reasons:
- logging on garbage collection is best-effort and may happen some time after
the error, if at all
- it can be hard to figure out where the error actually happened.
- it is logged as a scary CRITICAL error which (a) I always forget to grep for
and (b) it's not really CRITICAL if a background process we don't care about
fails.
So this is an attempt to add exception handling to everything we fire off into
the background.
This is mainly done by moving the calculation of where to send presence
updates from the presence handler to the transaction queue, so we only
need to send the presence event (and not the destinations) across the
replication connection. Before we were duplicating by sending the full
state across once per destination.
This fixes a class of 'Unexpected logcontext' messages, which were happening
because the logcontext was somewhat arbitrarily swapping between the sentinel
and the `run` logcontext.
Erik Johnston