This fixed the weirdness of 400 vs 404 as http status code in the case
the filter id is not known by the server.
As e.g. matrix-js-sdk expects 404 to catch this situation this leads
to unwanted behaviour.
Hopefully this will fix the occasional failures we were seeing in the room directory.
The problem was that events are not necessarily persisted (and `current_state_delta_stream` updated) in the same order as their stream_id. So for instance current_state_delta 9 might be persisted *before* current_state_delta 8. Then, when the room stats saw stream_id 9, it assumed it had done everything up to 9, and never came back to do stream_id 8.
We can solve this easily by only processing up to the stream_id where we know all events have been persisted.
Basically, if the federation sender starts getting behind, insert some sleeps
into the transaction transmission code to give the fed sender a chance to catch
up.
Might have to experiment a bit with the numbers.
It turns out that _local_membership_update doesn't run when you join a new, remote room. It only runs if you're joining a room that your server already knows about. This would explain #4703 and #5295 and why the transfer would work in testing and some rooms, but not others. This would especially hit single-user homeservers.
The check has been moved to right after the room has been joined, and works much more reliably. (Though it may still be a bit awkward of a place).
More often than not passing bytes to `txn.execute` is a bug (where we
meant to pass a string) that just happens to work if `BYTEA_OUTPUT` is
set to `ESCAPE`. However, this is a bit of a footgun so we want to
instead error when this happens, and force using `bytearray` if we
actually want to use bytes.
* Fix /federation/v1/state for recent room versions
Turns out this endpoint was completely broken for v3 rooms. Hopefully this
re-signing code is irrelevant nowadays anyway.
Erik Johnston