Currently when fetching state groups from the data store we make two
hits two the database: once for members and once for non-members (unless
request is filtered to one or the other). This adds needless load to the
datbase, so this PR refactors the lookup to make only a single database
hit.
Splits the state_group_cache in two.
One half contains normal state events; the other contains member events.
The idea is that the lazyloading common case of: "I want a subset of member events plus all of the other state" can be accomplished efficiently by splitting the cache into two, and asking for "all events" from the non-members cache, and "just these keys" from the members cache. This means we can avoid having to make DictionaryCache aware of these sort of complicated queries, whilst letting LL requests benefit from the caching.
Previously we were unable to sensibly use the caching and had to pull all state from the DB irrespective of the filtering, which made things slow. Hopefully fixes https://github.com/matrix-org/synapse/issues/3720.
This is the first tranche of support for room versioning. It includes:
* setting the default room version in the config file
* new room_version param on the createRoom API
* storing the version of newly-created rooms in the m.room.create event
* fishing the version of existing rooms out of the m.room.create event
When _get_state_for_groups is given a wildcard filter, just do a complete
lookup. Hopefully this will give us the best of both worlds by not filling up
the ram if we only need one or two keys, but also making the cache still work
for the federation reader usecase.
Fixes a regression that had crept in where the caching layer upholds requests for loading state which is filtered by type (but not by state_key), but the DB layer itself would interpret a missing state_key as a request to filter by null state_key rather than returning all state_keys.
* Split state group persist into seperate storage func
* Add per database engine code for state group id gen
* Move store_state_group to StateReadStore
This allows other workers to use it, and so resolve state.
* Hook up store_state_group
* Fix tests
* Rename _store_mult_state_groups_txn
* Rename StateGroupReadStore
* Remove redundant _have_persisted_state_group_txn
* Update comments
* Comment compute_event_context
* Set start val for state_group_id_seq
... otherwise we try to recreate old state groups
* Update comments
* Don't store state for outliers
* Update comment
* Update docstring as state groups are ints
... so that we don't need to secretly gut-wrench it for use in the slaved
stores. I haven't done the other stores yet, but we should. I'm tired of the
workers breaking every time we tweak the stores because I forgot to gut-wrench
the right method.
fixes https://github.com/matrix-org/synapse/issues/2655.
Add db_conn parameters to the `__init__` methods of the *Store classes, so that
they are all consistent, which makes the multiple inheritance work correctly
(and so that we can later extract mixins which can be used in the slavedstores)
Most of the time was spent copying a dict to filter out sentinel values
that indicated that keys did not exist in the dict. The sentinel values
were added to ensure that we cached the non-existence of keys.
By updating DictionaryCache to keep track of which keys were known to
not exist itself we can remove a dictionary copy.
Using _simple_select_list is fairly expensive for functions that return
a lot of rows and/or get called a lot. (This is because it carefully
constructs a list of dicts).
get_current_state_ids gets called a lot on startup and e.g. when the IRC
bridge decided to send tonnes of joins/leaves (as it invalidates the
cache). We therefore replace it with a custon txn function that builds
up the final result dict without building up and intermediate
representation.
* `get_forward_extremeties_for_room` takes a numeric `stream_ordering`. We were
passing a `RoomStreamToken`, which meant that it returned the *current*
extremities, rather than those corresponding to the `from_token`. However:
* `get_state_ids_for_events` required a second ('types') parameter; this meant
that a `TypeError` was thrown and we ended up acting as though there was *no*
prev state.
* `get_state_ids_for_events` actually returns a map from event_id to state
dictionary - just looking up the state keys in it again meant that we acted
as though there was no prev state. We now check if each member's state has
changed since *any* of the extremities.
Also add/fix some comments.
Will Hunt