Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>

Speed up processing of federation stream RDATA rows.

Instead of storing and sending an ACK for every single row we send
synchronously, we instead do it asynchronously while batching up
updates.

Signed-off-by: Christopher Cooper <cooperc@ocf.berkeley.edu>

Synapse 1.14.0rc2 (2020-05-27)
==============================

Bugfixes
--------

- Fix cache config to not apply cache factor to event cache. Regression in v1.14.0rc1. ([\#7578](https://github.com/matrix-org/synapse/issues/7578))
- Fix bug where `ReplicationStreamer` was not always started when replication was enabled. Bug introduced in v1.14.0rc1. ([\#7579](https://github.com/matrix-org/synapse/issues/7579))
- Fix specifying individual cache factors for caches with special characters in their name. Regression in v1.14.0rc1. ([\#7580](https://github.com/matrix-org/synapse/issues/7580))

Improved Documentation
----------------------

- Fix the OIDC `client_auth_method` value in the sample config. ([\#7581](https://github.com/matrix-org/synapse/issues/7581))

We were using `logger` syntax which isn't supported by `Exception`s.

'client_auth_method' commented out value was erronously 'client_auth_basic',
when code and docstring says it should be 'client_secret_basic'.

Signed-off-by: Jason Robinson <jasonr@matrix.org>

This mostly applise to `*stateGroupCache*` and co.

Broke in #6391.

This is already correctly done when we instansiate the cache, but wasn't
when it got reloaded (which always happens at least once on startup).

Fixes #7566.

Since this is not a full release yet

The bg update never managed to complete, because it kept being interrupted by
transactions which want to take a lock.

Just doing it in the foreground isn't that bad, and is a good deal simpler.

Introduced in #7556

A couple of changes of significance:

 * remove the `_last_ack < federation_position` condition, so that
   updates will still be correctly processed after restart

 * Correctly wire up send_federation_ack to the right class.

we can use `make_in_list_sql_clause` rather than doing our own half-baked
equivalent, which has the benefit of working just fine with empty lists.

(This has quite a lot of tests, so I think it's pretty safe)

These are surprisingly expensive, and we only really need to do them at startup.

Broke in #7534.

The idea here is that if an instance persists an event via the replication HTTP API it can return before we receive that event over replication, which can lead to races where code assumes that persisting an event immediately updates various caches (e.g. current state of the room).

Most of Synapse doesn't hit such races, so we don't do the waiting automagically, instead we do so where necessary to avoid unnecessary delays. We may decide to change our minds here if it turns out there are a lot of subtle races going on.

People probably want to look at this commit by commit.