Commit Graph

123 Commits

Author SHA1 Message Date
Erik Johnston 00f51493f5 Fix reindex 2016-09-14 10:18:30 +01:00
Erik Johnston d5ae1f1291 Ensure we don't mutate state cache entries 2016-09-14 10:03:48 +01:00
Erik Johnston 03a98aff3c Create new index concurrently 2016-09-12 14:27:01 +01:00
Erik Johnston 897d57bc58 Change state fetch query for postgres to be faster
It turns out that postgres doesn't like doing a list of OR's and is
about 1000x slower, so we just issue a query for each specific type
seperately.
2016-09-12 10:05:07 +01:00
Erik Johnston 5beda10bbd Reindex state_groups_state after pruning 2016-09-08 16:18:01 +01:00
Erik Johnston b568ca309c Temporarily disable sequential scans for state fetching 2016-09-08 09:38:54 +01:00
Erik Johnston 513188aa56 Comment 2016-09-07 14:53:23 +01:00
Erik Johnston fadb01551a Add appopriate framing clause 2016-09-07 14:39:01 +01:00
Erik Johnston d25c20ccbe Use windowing function to make use of index 2016-09-07 14:22:22 +01:00
Erik Johnston 0595413c0f Scale the batch size so that we're not bitten by the minimum 2016-09-05 15:49:57 +01:00
Erik Johnston a7032abb2e Correctly handle reindexing state groups that already have an edge 2016-09-05 15:07:23 +01:00
Erik Johnston 70332a12dd Take value in a better way 2016-09-05 14:57:14 +01:00
Erik Johnston 373654c635 Comment about sqlite and WITH RECURSIVE 2016-09-05 14:50:36 +01:00
Erik Johnston 628e65721b Add comments 2016-09-05 10:41:27 +01:00
Erik Johnston a99e933550 Add upgrade script that will slowly prune state_groups_state entries 2016-09-05 10:05:36 +01:00
Erik Johnston 598317927c Limit the length of state chains 2016-09-02 10:41:38 +01:00
Erik Johnston 9e25443db8 Move to storing state_groups_state as deltas 2016-09-01 14:31:26 +01:00
Erik Johnston 0cfd6c3161 Use state_groups table to test existence 2016-08-31 16:25:57 +01:00
Erik Johnston c10cb581c6 Correctly handle the difference between prev and current state 2016-08-31 14:26:22 +01:00
Erik Johnston 1bb8ec296d Generate state group ids in state layer 2016-08-31 10:09:46 +01:00
Erik Johnston 5dc2a702cf Make _state_groups_id_gen a normal IdGenerator 2016-08-30 16:55:11 +01:00
Erik Johnston 778fa85f47 Make sync not pull out full state 2016-08-25 18:59:44 +01:00
Erik Johnston a3dc1e9cbe Replace context.current_state with context.current_state_ids 2016-08-25 17:32:22 +01:00
Erik Johnston 17f4f14df7 Pull out event ids rather than full events for state 2016-08-25 13:42:44 +01:00
Erik Johnston ba214a5e32 Remove lru option 2016-08-19 14:17:11 +01:00
Erik Johnston 61c7edfd34 Add cache to _get_state_groups_from_groups 2016-04-19 17:22:03 +01:00
Mark Haines 87f2dec8d4 Make the cache objects be per instance rather than being global 2016-04-06 13:08:05 +01:00
Mark Haines 89e6839a48 Merge pull request #686 from matrix-org/markjh/doc_strings
Use google style doc strings.
2016-04-01 16:20:09 +01:00
Mark Haines 2a37467fa1 Use google style doc strings.
pycharm supports them so there is no need to use the other format.

Might as well convert the existing strings to reduce the risk of
people accidentally cargo culting the wrong doc string format.
2016-04-01 16:12:07 +01:00
Mark Haines e36bfbab38 Use a stream id generator for backfilled ids 2016-04-01 13:29:05 +01:00
Mark Haines 31a9eceda5 Add a replication stream for state groups 2016-03-30 16:01:58 +01:00
Mark Haines 1e25f62ee6 Use a stream id generator to assign state group ids 2016-03-30 12:55:02 +01:00
Erik Johnston 2f0180b09e Don't bother interning keys that are already interned 2016-03-23 16:29:46 +00:00
Erik Johnston acdfef7b14 Intern all the things 2016-03-23 16:25:54 +00:00
Erik Johnston 75daede92f String intern 2016-03-23 14:53:53 +00:00
Erik Johnston 8b0dfc9fc4 Don't cache events in get_current_state_for_key 2016-03-23 11:42:17 +00:00
Erik Johnston 2c86187a1b Don't cache events in _state_group_cache
Instead, simply cache the event ids, relying on the event cache to cache
the actual events.

The problem was that while the state groups cache was limited in the
number of groups it could hold, each individual group could consist of
thousands of events.
2016-03-22 12:00:09 +00:00
Mark Haines 54172924c8 Load the current id in the IdGenerator constructor
Rather than loading them lazily. This allows us to remove all
the yield statements and spurious arguments for the get_next
methods.

It also allows us to replace all instances of get_next_txn with
get_next since get_next no longer needs to access the db.
2016-03-01 14:32:56 +00:00
Erik Johnston 5189bfdef4 Batch fetch _get_state_groups_from_groups 2016-02-10 13:24:42 +00:00
Erik Johnston 24f00a6c33 Use _simple_select_many for _get_state_group_for_events 2016-02-10 12:57:50 +00:00
Matthew Hodgson 6c28ac260c copyrights 2016-01-07 04:26:29 +00:00
Richard van der Hoff fddedd51d9 Fix a few race conditions in the state calculation
Be a bit more careful about how we calculate the state to be returned by
/sync. In a few places, it was possible for /sync to return slightly later
state than that represented by the next_batch token and the timeline. In
particular, the following cases were susceptible:

* On a full state sync, for an active room
* During a per-room incremental sync with a timeline gap
* When the user has just joined a room. (Refactor check_joined_room to make it
  less magical)

Also, use store.get_state_for_events() (and thus the existing stategroups) to
calculate the state corresponding to a particular sync position, rather than
state_handler.compute_event_context(), which recalculates from first principles
(and tends to miss some state).

Merged from PR https://github.com/matrix-org/synapse/pull/372
2015-11-13 10:39:09 +00:00
Erik Johnston 927004e349 Remove unused room_id parameter 2015-10-12 15:06:14 +01:00
Mark Haines 1cd65a8d1e synapse/storage/state.py: _make_group_id was unused 2015-09-23 10:37:58 +01:00
Erik Johnston 1bd1a43073 Actually check if event_id isn't returned by _get_state_groups 2015-08-21 14:30:34 +01:00
Erik Johnston a82938416d Remove newline because vertical whitespace makes mjark sad 2015-08-18 16:28:13 +01:00
Erik Johnston 0bfdaf1f4f Rejig the code to make it nicer 2015-08-18 16:26:07 +01:00
Erik Johnston 8199475ce0 Ensure we never return a None event from _get_state_for_groups 2015-08-18 11:44:10 +01:00
Erik Johnston 85d0bc3bdc Reduce cache size from obscenely large to quite large 2015-08-18 11:00:38 +01:00
Erik Johnston f9d4da7f45 Fix bug where we were leaking None into state event lists 2015-08-17 09:39:45 +01:00