Commit Graph

379 Commits

Author SHA1 Message Date
Erik Johnston
8f65ab98d2 Remove unnecessary iteritems 2018-07-24 15:08:01 +01:00
Erik Johnston
ed0dd68731 Fixup comment (and indent) 2018-07-24 14:31:38 +01:00
Erik Johnston
811ac73a42 Don't fetch state from the database unless needed 2018-07-24 14:18:23 +01:00
Erik Johnston
a79410e7b8 Have _get_new_state_after_events return delta
If we have a delta from the existing to new current state, then we can
reuse that rather than manually working it out by fetching both lots of
state.
2018-07-24 14:14:17 +01:00
Erik Johnston
223341205e Don't require to_delete to have event_ids 2018-07-24 14:02:40 +01:00
Erik Johnston
97acd385a3 Merge branch 'develop' of github.com:matrix-org/synapse into erikj/speed_up_calculate_state_delta 2018-07-24 11:32:13 +01:00
Erik Johnston
0fa73e4a63 Remove unnecessary if 2018-07-24 11:19:23 +01:00
Erik Johnston
ff5426f6b8 Speed up _calculate_state_delta 2018-07-24 10:55:11 +01:00
Richard van der Hoff
a678145010
Merge branch 'develop' into rav/logcontext_fixes 2018-07-24 10:43:30 +01:00
Richard van der Hoff
8dff6e0322 Logcontext fixes
Fix some random logcontext leaks.
2018-07-24 00:37:17 +01:00
Richard van der Hoff
69fb5dbdab fix idiocy 2018-07-24 00:04:44 +01:00
Richard van der Hoff
1938cffaea Add some measure blocks to persist_events
... to help us figure out where 40% of CPU is going
2018-07-23 23:48:19 +01:00
Erik Johnston
8b9f164fff Comments 2018-07-23 17:43:01 +01:00
Erik Johnston
50c60e5fad Only get cached state from context in persist_event
We don't want to bother pulling out the current state from the DB since
until we know we have to. Checking the context for state is just an
optimisation.
2018-07-23 17:21:40 +01:00
Erik Johnston
e42510ba63 Use new getters 2018-07-23 13:17:22 +01:00
Richard van der Hoff
667fba68f3 Run things as background processes
This fixes #3518, and ensures that we get useful logs and metrics for lots of
things that happen in the background.

(There are certainly more things that happen in the background; these are just
the common ones I've found running a single-process synapse locally).
2018-07-18 20:55:05 +01:00
Amber Brown
49af402019 run isort 2018-07-09 16:09:20 +10:00
Erik Johnston
3905c693c5 Invalidate cache on correct thread 2018-07-02 11:36:44 +01:00
Amber Brown
6350bf925e
Attempt to be more performant on PyPy (#3462) 2018-06-28 14:49:57 +01:00
Mark Haines
9850f66abe Deleting from event_push_actions needs to use an index 2018-06-22 15:54:48 +01:00
Amber Brown
c936a52a9e
Consistently use six's iteritems and wrap lazy keys/values in list() if they're not meant to be lazy (#3307) 2018-05-31 19:03:47 +10:00
Amber Brown
53cc2cde1f cleanup 2018-05-22 17:32:57 -05:00
Amber Brown
a8990fa2ec Merge remote-tracking branch 'origin/develop' into 3218-official-prom 2018-05-22 10:50:26 -05:00
Amber Brown
fcc525b0b7 rest of the changes 2018-05-21 19:48:57 -05:00
Erik Johnston
680530cc7f Clarify comment 2018-05-16 11:47:29 +01:00
Erik Johnston
43e6e82c4d Comments 2018-05-16 11:13:31 +01:00
Erik Johnston
c945af8799 Move and rename variable 2018-05-16 10:52:06 +01:00
Erik Johnston
5f27ed75ad Make purge_history operate on tokens
As we're soon going to change how topological_ordering works
2018-05-15 16:23:50 +01:00
Erik Johnston
37dbee6490 Use events_to_purge table rather than token 2018-05-15 16:23:47 +01:00
Erik Johnston
31c7c29d43 Fix up grammar 2018-05-03 10:38:58 +01:00
Erik Johnston
6e10eed28e Refactor event storage to not require state
This is in preparation for using contexts that may or may not have the
current_state_ids set. This will allow us to avoid unnecessarily pulling
out state for an event on the master process when using workers.

We also add a check to see if the state groups of the old extremities
are the same as the new ones.
2018-04-27 11:38:02 +01:00
Richard van der Hoff
b1dfbc3c40 Refactor store.have_events
It turns out that most of the time we were calling have_events, we were only
using half of the result. Replace have_events with have_seen_events and
get_rejection_reasons, so that we can see what's going on a bit more clearly.
2018-04-20 10:25:56 +01:00
Erik Johnston
92e34615c5 Track where event stream processing have gotten up to 2018-04-11 12:13:40 +01:00
Richard van der Hoff
05630758f2 Use static JSONEncoders
using json.dumps with custom options requires us to create a new JSONEncoder on
each call. It's more efficient to create one upfront and reuse it.
2018-03-29 23:13:33 +01:00
Erik Johnston
800cfd5774 Comment 2018-03-27 13:30:39 +01:00
Erik Johnston
152c2ac19e Fix indent 2018-03-27 13:13:46 +01:00
Erik Johnston
e70287cff3 Comment 2018-03-27 13:13:38 +01:00
Erik Johnston
3f49e131d9 Add counter metrics for calculating state delta
This will allow us to measure how often we calculate state deltas in
event persistence that we would have been able to calculate at the same
time we calculated the state for the event.
2018-03-27 10:57:35 +01:00
Erik Johnston
fa72803490 Merge branch 'master' of github.com:matrix-org/synapse into develop 2018-03-19 11:41:01 +00:00
Erik Johnston
a8ce159be4 Replace some ujson with simplejson to make it work 2018-03-16 00:27:09 +00:00
Erik Johnston
5b631ff41a Remove wrong comment 2018-03-16 00:07:08 +00:00
Erik Johnston
926ba76e23 Replace ujson with simplejson 2018-03-15 23:43:31 +00:00
Richard van der Hoff
889a2a853a Add Measure block for persist_events
This seems like a useful thing to measure.
2018-03-13 10:01:42 +00:00
Erik Johnston
8cb44da4aa Fix race in sync when joining room
The race happens when the user joins a room at the same time as doing a
sync. We fetch the current token and then get the rooms the user is in.
If the join happens after the current token, but before we get the rooms
we end up sending down a partial room entry in the sync.

This is fixed by looking at the stream ordering of the membership
returned by get_rooms_for_user, and handling the case when that stream
ordering is after the current token.
2018-03-07 11:55:31 +00:00
Erik Johnston
33bebb63f3 Add some caches to help read marker API 2018-03-01 17:08:17 +00:00
Erik Johnston
45b5fe9122 Merge branch 'develop' of github.com:matrix-org/synapse into erikj/handle_unpersisted_events_push 2018-02-26 13:49:24 +00:00
Erik Johnston
eba93b05bf Split EventsWorkerStore into separate file 2018-02-23 11:01:21 +00:00
Erik Johnston
bf8a36e080 Update copyright 2018-02-23 10:52:10 +00:00
Erik Johnston
5d0f665848 Remove redundant clock 2018-02-23 10:49:58 +00:00
Erik Johnston
3bd760628b _event_persist_queue shouldn't be in worker store 2018-02-23 10:49:18 +00:00
Erik Johnston
27b094f382 Split out get_events and co into a worker store 2018-02-21 11:41:48 +00:00
Erik Johnston
24087bffa9 Ensure all push actions are deleted from staging 2018-02-20 12:34:31 +00:00
Erik Johnston
ad0ccf15ea Refactor _set_push_actions_for_event_and_users_txn to use events_and_contexts 2018-02-20 12:34:28 +00:00
Richard van der Hoff
d1a3325f99 (Really) fix tablescan of event_push_actions on purge
commit 278d21b5 added new code to avoid the tablescan, but didn't remove the
old :/
2018-02-16 14:02:31 +00:00
Erik Johnston
4810f7effd Remove context.push_actions 2018-02-15 15:47:06 +00:00
Erik Johnston
c714c61853 Update event_push_actions table from staging table 2018-02-15 15:47:06 +00:00
Richard van der Hoff
39bf47319f purge_history: fix sqlite syntax error
apparently sqlite insists on indexes being named
2018-02-14 16:42:19 +00:00
Richard van der Hoff
ac27f6a35e purge_history: handle sqlite asshattery
apparently creating a temporary table commits the transaction. because that's a
useful thing.
2018-02-14 16:41:12 +00:00
Richard van der Hoff
278d21b5e4 purge_history: fix index use
event_push_actions doesn't have an index on event_id, so we need to specify
room_id.
2018-02-14 15:44:51 +00:00
Richard van der Hoff
5fcbf1e07c Rework event purge to use a temporary table
... which should speed things up by reducing the amount of data being shuffled
across the connection
2018-02-14 11:02:22 +00:00
Richard van der Hoff
d627174da2 Fix log message in purge_history
(we don't just remove remote events)
2018-02-13 16:51:21 +00:00
Richard van der Hoff
bfdf7b9237
Merge pull request #2864 from matrix-org/rav/persist_event_caching
Use StateResolutionHandler to resolve state in persist_events
2018-02-13 14:45:57 +00:00
Richard van der Hoff
630caf8a70
style nit 2018-02-13 14:29:22 +00:00
Richard van der Hoff
10b34dbb9a
Merge pull request #2858 from matrix-org/rav/purge_updates
delete_local_events for purge_room_history
2018-02-09 14:11:00 +00:00
Richard van der Hoff
39a6b35496 purge: move room_depth update to end
... to avoid locking the table for too long
2018-02-09 13:07:41 +00:00
Richard van der Hoff
74fcbf741b delete_local_events for purge_history
Add a flag which makes the purger delete local events
2018-02-09 13:07:41 +00:00
Richard van der Hoff
e571aef06d purge: Move cache invalidation to more appropriate place
it was a bit of a non-sequitur there
2018-02-09 13:07:41 +00:00
Richard van der Hoff
61ffaa8137 bump purge logging to info
this thing takes ages and the only sign of any progress is the logs, so having
some logs is useful.
2018-02-09 13:07:41 +00:00
Richard van der Hoff
671540dccf rename delete_old_state -> purge_history
(beacause it deletes more than state)
2018-02-09 13:07:41 +00:00
Erik Johnston
3d33eef6fc
Store state groups separately from events (#2784)
* 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
2018-02-06 14:31:24 +00:00
Richard van der Hoff
ebfe64e3d6 Use StateResolutionHandler to resolve state in persist events
... and thus benefit (hopefully) from its cache.
2018-02-05 16:23:26 +00:00
Richard van der Hoff
225dc3b4cb Flatten _get_new_state_after_events
rejig the if statements to simplify the logic and reduce indentation
2018-02-05 16:23:25 +00:00
Richard van der Hoff
9fcbbe8e7d Check that events being persisted have state_group 2018-02-05 16:23:25 +00:00
Richard van der Hoff
447aed42d2 Add event_map param to resolve_state_groups 2018-02-05 16:23:25 +00:00
Richard van der Hoff
ee6fb4cf85 Remove redundant return value from _calculate_state_delta
we already have the state from _get_new_state_after_events, so returning it
from _calculate_state_delta is just confusing.
2018-02-05 16:23:20 +00:00
Richard van der Hoff
e1e4ec9f9d factor _get_new_state_after_events out of _calculate_state_delta
This reduces the scope of a bunch of variables
2018-01-31 21:32:09 +00:00
Richard van der Hoff
b387ee17b6 Improve exception handling in persist_event
1. use `deferred.errback()` instead of `deferred.errback(e)`, which means that
a Failure object will be constructed using the current exception state,
*including* its stack trace - so the stack trace is saved in the Failure,
leading to better exception reports.

2. Set `consumeErrors=True` on the ObservableDeferred, because we know that
there will always be at least one observer - which avoids a spurious "CRITICAL:
unhandled exception in Deferred" error in the logs
2018-01-29 17:05:33 +00:00
Richard van der Hoff
36da256cc6
Merge pull request #2805 from matrix-org/rav/log_state_res
Log room when doing state resolution
2018-01-17 18:05:04 +00:00
Richard van der Hoff
1224612a79 Log room when doing state resolution
Mostly because it helps figure out what is prompting the resolution
2018-01-17 17:11:59 +00:00
Richard van der Hoff
390093d45e Split resolve_events into two functions
... so that the return type doesn't depend on the arg types
2018-01-17 15:44:31 +00:00
Richard van der Hoff
64ddec1bc0 Fix a logcontext leak in persist_events
ObserveableDeferred expects its callbacks to be called without any
logcontexts, whereas it turns out we were calling them with the logcontext of
the request which initiated the persistence loop.

It seems wrong that we are attributing work done in the persistence loop to the
request that happened to initiate it, so let's solve this by dropping the
logcontext for it.

(I'm not sure this actually causes any real problems other than messages in the
debug log, but let's clean it up anyway)
2018-01-16 11:47:36 +00:00
Richard van der Hoff
6cfee09be9 Make __init__ consitstent across Store heirarchy
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)
2017-11-13 10:46:07 +00:00
Richard van der Hoff
eaaabc6c4f replace 'except:' with 'except Exception:'
what could possibly go wrong
2017-10-23 15:52:32 +01:00
Richard van der Hoff
2e9f5ea31a Fix logcontext handling for persist_events
* don't use preserve_context_over_deferred, which is known broken.

* remove a redundant preserve_fn.

* add/improve some comments
2017-10-17 10:59:30 +01:00
Erik Johnston
11d62f43c9 Invalidate cache 2017-10-03 14:12:28 +01:00
Erik Johnston
6ff14ddd2e Make into list 2017-06-29 15:47:37 +01:00
Erik Johnston
5946aa0877 Prefill forward extrems and event to state groups 2017-06-29 15:38:48 +01:00
Erik Johnston
dfeca6cf40 Merge pull request #2286 from matrix-org/erikj/split_out_user_dir
Split out user directory to a separate process
2017-06-16 13:01:19 +01:00
Erik Johnston
6aa5bc8635 Initial worker impl 2017-06-16 11:47:11 +01:00
Erik Johnston
052c5d19d5 Merge pull request #2281 from matrix-org/erikj/phone_home_stats
Fix phone home stats
2017-06-15 12:46:23 +01:00
Erik Johnston
4b461a6931 Add some more stats 2017-06-15 09:39:39 +01:00
Erik Johnston
617304b2cf Fix phone home stats 2017-06-14 19:47:15 +01:00
Erik Johnston
d53fe399eb Add cache for is_host_joined 2017-06-13 09:56:18 +01:00
Erik Johnston
c62c480dc6 Merge pull request #2259 from matrix-org/erikj/fix_state_woes
Fix bug where state_group tables got corrupted
2017-06-07 17:51:25 +01:00
Erik Johnston
197bd126f0 Fix bug where state_group tables got corrupted
This is due to the fact that we prefilled caches using txn.call_after,
which always gets called including on error.

We fix this by making txn.call_after only fire when a transaction
completes successfully, which is what we want most of the time anyway.
2017-06-07 17:39:36 +01:00
Erik Johnston
4abcff0177 Fix typo 2017-05-31 16:22:36 +01:00
Erik Johnston
304880d185 Add stream change cache 2017-05-31 15:46:36 +01:00
Erik Johnston
04095f7581 Add clobbered event_id 2017-05-30 14:53:01 +01:00
Erik Johnston
a584a81b3e Add current_state_delta_stream table 2017-05-30 14:44:09 +01:00
Erik Johnston
c049472b8a Only store event_auth for state events 2017-05-24 15:23:31 +01:00