Commit Graph

108 Commits

Author SHA1 Message Date
PiotrKozimor
12649ccedd
Improve selectRoomIDsWithAnyMembershipSQL performance (#2738)
Recently I have observed that dendrite spends a lot of time (~390s) in
`selectRoomIDsWithAnyMembershipSQL` query

```
dendrite_syncapi=# select total_exec_time, left(query,100) from pg_stat_statements order by total_exec_time desc limit 5 ;
  total_exec_time   |                                                 left
--------------------+------------------------------------------------------------------------------------------------------
  747826.5800519128 | SELECT event_id, id, headered_event_json, session_id, exclude_from_sync, transaction_id, history_vis
  389130.5490339942 | SELECT DISTINCT room_id, membership FROM syncapi_current_room_state WHERE type = $2 AND state_key =
 376104.17514700035 | SELECT psd.datname, xact_commit, xact_rollback, blks_read, blks_hit, tup_returned, tup_fetched, tup_
   363644.164092031 | SELECT event_type_nid, event_state_key_nid, event_nid FROM roomserver_events WHERE event_nid = ANY($
  58570.48104699995 | SELECT event_id, headered_event_json FROM syncapi_current_room_state WHERE room_id = $1 AND ( $2::te
(5 rows)
```

Explain analyze showed correct usage of `syncapi_room_state_unique`
index:

```
dendrite_syncapi=#
explain analyze SELECT distinct room_id, membership FROM syncapi_current_room_state WHERE type = 'm.room.member' AND state_key = '@qjfl:dendrite.stg.globekeeper.com';
                                                                               QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 Unique  (cost=2749.38..2749.56 rows=24 width=52) (actual time=2.933..2.956 rows=65 loops=1)
   ->  Sort  (cost=2749.38..2749.44 rows=24 width=52) (actual time=2.932..2.937 rows=65 loops=1)
         Sort Key: room_id, membership
         Sort Method: quicksort  Memory: 34kB
         ->  Index Scan using syncapi_room_state_unique on syncapi_current_room_state  (cost=0.41..2748.83 rows=24 width=52) (actual time=0.030..2.890 rows=65 loops=1)
               Index Cond: ((type = 'm.room.member'::text) AND (state_key = '@qjfl:dendrite.stg.globekeeper.com'::text))
 Planning Time: 0.140 ms
 Execution Time: 2.990 ms
(8 rows)
```

Multi-column indexes in Postgres shall perform well for leftmost
columns, but I gave it a try and created
`syncapi_current_room_state_type_state_key_idx` index. I could observe
significant performance improvement. Execution time dropped from 2.9 ms
to 0.24 ms:

```
explain analyze SELECT distinct room_id, membership FROM syncapi_current_room_state WHERE type = 'm.room.member' AND state_key = '@qjfl:dendrite.stg.globekeeper.com';
                                                                             QUERY PLAN
--------------------------------------------------------------------------------------------------------------------------------------------------------------------
 Unique  (cost=96.46..96.64 rows=24 width=52) (actual time=0.199..0.218 rows=65 loops=1)
   ->  Sort  (cost=96.46..96.52 rows=24 width=52) (actual time=0.199..0.202 rows=65 loops=1)
         Sort Key: room_id, membership
         Sort Method: quicksort  Memory: 34kB
         ->  Bitmap Heap Scan on syncapi_current_room_state  (cost=4.53..95.91 rows=24 width=52) (actual time=0.048..0.139 rows=65 loops=1)
               Recheck Cond: ((type = 'm.room.member'::text) AND (state_key = '@qjfl:dendrite.stg.globekeeper.com'::text))
               Heap Blocks: exact=59
               ->  Bitmap Index Scan on syncapi_current_room_state_type_state_key_idx  (cost=0.00..4.53 rows=24 width=0) (actual time=0.037..0.037 rows=65 loops=1)
                     Index Cond: ((type = 'm.room.member'::text) AND (state_key = '@qjfl:dendrite.stg.globekeeper.com'::text))
 Planning Time: 0.236 ms
 Execution Time: 0.242 ms
(11 rows)
```

Next improvement is skipping DISTINCT and rely on map assignment in
`SelectRoomIDsWithAnyMembership`. Execution time drops by almost half:

```
explain analyze SELECT room_id, membership FROM syncapi_current_room_state WHERE type = 'm.room.member' AND state_key = '@qjfl:dendrite.stg.globekeeper.com';
                                                                       QUERY PLAN
--------------------------------------------------------------------------------------------------------------------------------------------------------
 Bitmap Heap Scan on syncapi_current_room_state  (cost=4.53..95.91 rows=24 width=52) (actual time=0.032..0.113 rows=65 loops=1)
   Recheck Cond: ((type = 'm.room.member'::text) AND (state_key = '@qjfl:dendrite.stg.globekeeper.com'::text))
   Heap Blocks: exact=59
   ->  Bitmap Index Scan on syncapi_current_room_state_type_state_key_idx  (cost=0.00..4.53 rows=24 width=0) (actual time=0.021..0.021 rows=65 loops=1)
         Index Cond: ((type = 'm.room.member'::text) AND (state_key = '@qjfl:dendrite.stg.globekeeper.com'::text))
 Planning Time: 0.087 ms
 Execution Time: 0.136 ms
(7 rows)
```

In our env we spend only 1s on inserting to table, so the write penalty
of creating an index should be small.
```
dendrite_syncapi=# select total_exec_time, left(query,100) from pg_stat_statements where query like '%INSERT%syncapi_current_room_state%' order by total_exec_time desc;
  total_exec_time   |                                                 left
--------------------+------------------------------------------------------------------------------------------------------
 1139.9057619999971 | INSERT INTO syncapi_current_room_state (room_id, event_id, type, sender, contains_url, state_key, he
(1 row)
``` 

This PR does not require test modifications.

### Pull Request Checklist

<!-- Please read docs/CONTRIBUTING.md before submitting your pull
request -->

* [x] I have added added tests for PR _or_ I have justified why this PR
doesn't need tests.
* [x] Pull request includes a [sign
off](https://github.com/matrix-org/dendrite/blob/main/docs/CONTRIBUTING.md#sign-off)

Signed-off-by: `Piotr Kozimor <p1996k@gmail.com>`
2022-09-27 09:41:36 +01:00
Neil Alexander
97d7cf2232
Remove deleted state logging lines from sync API (they are pointless) 2022-09-20 11:25:18 +01:00
Neil Alexander
955e69a3b7
Optimise SharedUsers again by using complete composite index 2022-09-09 14:18:45 +01:00
Neil Alexander
6ee758df63
Optimise shared users query in Synx API slightly by removing a potential sort 2022-09-09 13:50:50 +01:00
Neil Alexander
646de03d60
More writer fixes in the Sync API 2022-09-09 13:06:42 +01:00
PiotrKozimor
2be43560ca
Index on syncapi_send_to_device table (#2684)
Introduced index improves select query performance. Example execution time of `selectSendToDeviceMessagesSQL` query dropped from 80 ms to 15 ms. No sytest modifications are required.

### Pull Request Checklist

* [x] I have added added tests for PR _or_ I have justified why this PR doesn't need tests.
* [x] Pull request includes a [sign off](https://github.com/matrix-org/dendrite/blob/main/docs/CONTRIBUTING.md#sign-off)

Signed-off-by: `Piotr Kozimor <p1996k@gmail.com>`
2022-08-30 14:47:54 +01:00
Neil Alexander
ad4ac2c016
Stop spamming the logs with StateBetween: ignoring deleted state event IDs 2022-08-16 14:42:35 +01:00
Till
05cafbd197
Implement history visibility on /messages, /context, /sync (#2511)
* Add possibility to set history_visibility and user AccountType

* Add new DB queries

* Add actual history_visibility changes for /messages

* Add passing tests

* Extract check function

* Cleanup

* Cleanup

* Fix build on 386

* Move ApplyHistoryVisibilityFilter to internal

* Move queries to topology table

* Add filtering to /sync and /context
Some cleanup

* Add passing tests; Remove failing tests :(

* Re-add passing tests

* Move filtering to own function to avoid duplication

* Re-add passing test

* Use newly added GMSL HistoryVisibility

* Update gomatrixserverlib

* Set the visibility when creating events

* Default to shared history visibility

* Remove unused query

* Update history visibility checks to use gmsl
Update tests

* Remove unused statement

* Update migrations to set "correct" history visibility

* Add method to fetch the membership at a given event

* Tweaks and logging

* Use actual internal rsAPI, default to shared visibility in tests

* Revert "Move queries to topology table"

This reverts commit 4f0d41be9c194a46379796435ce73e79203edbd6.

* Remove noise/unneeded code

* More cleanup

* Try to optimize database requests

* Fix imports

* PR peview fixes/changes

* Move setting history visibility to own migration, be more restrictive

* Fix unit tests

* Lint

* Fix missing entries

* Tweaks for incremental syncs

* Adapt generic changes

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
Co-authored-by: kegsay <kegan@matrix.org>
2022-08-11 18:23:35 +02:00
Till
cecd11be9a
Partly fix notification counts (#2621)
* Fix notification query

* Also for SQLite

* Move tests to whitelist

* Revert "Move tests to whitelist"

This reverts commit a7d0120019a111ce45a447ba40233d9c101e6e9b.
2022-08-05 13:44:20 +02:00
Till
9fe509b18d
Fix syncapi shared users query & device lists (#2614)
* Fix query issue, only add "changed" users if we actually share a room

* Avoid log spam if context is done

* Undo changes to filterSharedUsers

* Add logging again..

* Fix SQLite shared users query

* Change query to include invited users
2022-08-03 18:35:17 +02:00
Till
df5d4dc7a3
Delete correct Send-to-Device messages (#2608)
* Add send-to-device tests

* Update tests, fix message deletion

* PR comments
2022-08-02 17:00:16 +02:00
Till
081f5e7226
Update database migrations, remove goose (#2264)
* Add new db migration

* Update migrations
Remove goose

* Add possibility to test direct upgrades

* Try to fix WASM test

* Add checks for specific migrations

* Remove AddMigration
Use WithTransaction
Add Dendrite version to table

* Fix linter issues

* Update tests

* Update comments, outdent if

* Namespace migrations

* Add direct upgrade tests, skipping over one version

* Split migrations

* Update go version in CI

* Fix copy&paste mistake

* Use contexts in migrations

Co-authored-by: kegsay <kegan@matrix.org>
Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-07-25 10:39:22 +01:00
Till
a7e92f8cb9
History visibility database changes (#2533)
* Add new history_visibility column

* Update SQL queries to include history_visibility

* Store the history visibilty calculated by the roomserver

* Update GMSL

* Update migrations

* Fix migration

* Update GMSL

* Fix `go.sum`

* Update GMSL to use sql.Scanner & sql.Valuer

* Re-order migration/table creation

* Update gomatrixserverlib

* Add history_visibility column to current_room_state

* Fix migrations

* Return error instead of Fatal log

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-07-18 14:46:15 +02:00
Neil Alexander
90bf01d8b1
Use sync API database in filterSharedUsers (#2572)
* Add function to the sync API storage package for filtering shared users

* Use the database instead of asking the RS API

* Fix unit tests

* Fix map handling in `filterSharedUsers`
2022-07-15 16:25:26 +01:00
Till
09f0ff14c8
Minor SendToDevice fix (#2565)
* Avoid unnecessary marshalling if sending to the local server

* Fix ordering of ToDevice messages

* Revive SendToDevice test
2022-07-12 08:23:58 +02:00
Neil Alexander
e2a932ec0b
Add indexes to syncapi_output_room_events table that satisfy the filters (#2446) 2022-05-10 11:23:36 +01:00
Neil Alexander
4ad5f9c982
Global database connection pool (for monolith mode) (#2411)
* Allow monolith components to share a single database pool

* Don't yell about missing connection strings

* Rename field

* Setup tweaks

* Fix panic

* Improve configuration checks

* Update config

* Fix lint errors

* Update comments
2022-05-03 16:35:06 +01:00
Till
2a5b8e0306
Only load members of newly joined rooms (#2389)
* Only load members of newly joined rooms

* Comment that the query is prepared at runtime

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-04-28 18:53:28 +02:00
Till
21ee5b36a4
Limit presence in /sync responses (#2394)
* Use filter and limit presence count

* More limiting

* More limiting

* Fix unit test

* Also limit presence by last_active_ts

* Update query, use "from" as the initial lastPos

* Get 1000 presence events, they are filtered later

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-04-28 15:12:40 +01:00
Neil Alexander
655ac3e8fb
Try that again 2022-04-27 15:10:15 +01:00
Neil Alexander
6ee8507955
Correct account data position mapping 2022-04-27 15:10:10 +01:00
Neil Alexander
66b397b3c6
Don't create fictitious presence entries (#2381)
* Don't create fictitious presence entries for users that don't have any

* Update whitelist, since that test probably shouldn't be passing

* Fix panics
2022-04-27 11:25:07 +01:00
Neil Alexander
b527e33c16
Send all account data on complete sync by default
Squashed commit of the following:

commit 0ec8de57261d573a5f88577aa9d7a1174d3999b9
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Apr 26 16:56:30 2022 +0100

    Select filter onto provided target filter

commit da40b6fffbf5737864b223f49900048f557941f9
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Apr 26 16:48:00 2022 +0100

    Specify other field too

commit ffc0b0801f63bb4d3061b6813e3ce5f3b4c8fbcb
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Apr 26 16:45:44 2022 +0100

    Send as much account data as possible during complete sync
2022-04-26 16:58:20 +01:00
Neil Alexander
f6d07768a8
Fix account data position 2022-04-26 16:08:01 +01:00
Neil Alexander
6892e0f0e0
Start account data ID from from 2022-04-26 16:02:21 +01:00
Till
4c19f22725
Fix account_data not correctly send in a complete sync (#2379)
* Return the StreamPosition from the database and not the latest

* Fix linter issue
2022-04-26 14:50:56 +01:00
Till
e8be2b234f
Add heroes to the room summary (#2373)
* Implement room summary heroes

* Add passing tests

* Move MembershipCount to addRoomSummary

* Add comments, close Statement
2022-04-26 10:53:17 +02:00
Till
29f2168789
Make /messages filterable (#2347)
* Make /messages filterable
Fix bug when determining if an event contains an URL

* Add newly passing test

* Fix test
2022-04-13 13:16:02 +02:00
Till
69f2ff7c82
Correctly use provided filters (#2339)
* Apply filters correctly

* Fix issues; Use prepareWithFilters

* Update gmsl & tests

* go.mod..

* PR comments
2022-04-11 09:05:23 +02:00
kegsay
b4b2fbc36b Remove dead code in the sync api (#2341) 2022-04-09 00:37:50 +01:00
kegsay
6d25bd6ca5
syncapi: add more tests; fix more bugs (#2338)
* syncapi: add more tests; fix more bugs

bugfixes:
 - The postgres impl of TopologyTable.SelectEventIDsInRange did not use the provided txn
 - The postgres impl of EventsTable.SelectEvents did not preserve the ordering of the input event IDs in the output events slice
 - The sqlite impl of EventsTable.SelectEvents did not use a bulk `IN ($1)` query.

Added tests:
 - `TestGetEventsInRangeWithTopologyToken`
 - `TestOutputRoomEventsTable`
 - `TestTopologyTable`

* -p 1 for now
2022-04-08 17:53:24 +01:00
Till
60ee7eef4c
Add possibility to ignore users (#2329)
* Add ignore users

* Ignore users in pushrules
Add passing tests

* Update sytest lists

* Store ignore knowledge in the sync API

* Fix copyrights

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-04-07 15:08:19 +01:00
Till
e5e3350ce1
Add presence module V2 (#2312)
* Syncapi presence

* Clientapi http presence handler

* Why is this here?

* Missing files

* FederationAPI presence implementation

* Add new presence stream

* Pinecone update

* Pinecone update

* Add passing tests

* Make linter happy

* Add presence producer

* Add presence config option

* Set user to unavailable after x minutes

* Only set currently_active if online
Avoid unneeded presence updates when syncing

* Tweaks

* Query devices for last_active_ts
Fixes & tweaks

* Export SharedUsers/SharedUsers

* Presence stream in MemoryStorage

* Remove status_msg_nil

* Fix sytest crashes

* Make presence types const and use stringer for it

* Change options to allow inbound/outbound presence

* Fix option & typo

* Update configs

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-04-06 13:11:19 +02:00
Neil Alexander
cd8fac152e
Include joined and invite member counts in room summary (#2315)
* Include joined and invite member counts in room summary

This should fix #2314 and also fix the problem where some clients like Element Android, Fluffychat etc would display the wrong member count for a given room.

* Improve SQLite query precision

* Check existence of state key for membership events
2022-04-01 16:14:38 +01:00
S7evinK
49dc49b232
Remove eduserver (#2306)
* Move receipt sending to own JetStream producer

* Move SendToDevice to producer

* Remove most parts of the EDU server

* Fix SendToDevice & copyrights

* Move structs, cleanup EDU Server traces

* Use HeadersOnly subscription

* Missing file

* Fix linter issues

* Move consumers to own files

* Rename durable consumer; Consumer cleanup

* Docs/config cleanup
2022-03-29 14:14:35 +02:00
Neil Alexander
d983d17355
Fix lint errors 2022-03-24 10:03:22 +00:00
S7evinK
8336ce972e
Remove unused partition_offset_table (#2288) 2022-03-21 10:47:41 +01:00
Neil Alexander
475d3c1af9
Better mapping of stream positions to topological positions in /messages (#2263)
* Convert stream positions into topological positions for both `from` and `to` in `/messages`

* Hopefully it works now

* Remove unnecessary logging

* Return sane values if `StreamToTopologicalPosition` can't work out the right thing to do

* Revert logging change

* tweaks

* Fix `selectEventIDsInRangeASCSQL`

* Test `Getting messages going forward is limited for a departed room (SPEC-216)` was passing incorrectly so un-whitelist it
2022-03-18 10:40:01 +00:00
Neil Alexander
4e64c270db
Various bug fixes and tweaks around invites and membership 2022-03-17 17:05:21 +00:00
S7evinK
d8facd6308
Fix SQL statement for PurgeRoomState (#2280) 2022-03-16 11:25:50 +01:00
Neil Alexander
507a8e6773
Don't range entire state for /sync (#2270)
* Don't range entire state for rooms the user has no reason to care about

* Remove unnecessary db field in postgresql
2022-03-11 12:48:45 +00:00
Dan
f05ce478f0
Implement Push Notifications (#1842)
* Add Pushserver component with Pushers API

Co-authored-by: Tommie Gannert <tommie@gannert.se>
Co-authored-by: Dan Peleg <dan@globekeeper.com>

* Wire Pushserver component

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>

* Add PushGatewayClient.

The full event format is required for Sytest.

* Add a pushrules module.

* Change user API account creation to use the new pushrules module's defaults.

Introduces "scope" as required by client API, and some small field
tweaks to make some 61push Sytests pass.

* Add push rules query/put API in Pushserver.

This manipulates account data over User API, and fires sync messages
for changes. Those sync messages should, according to an existing TODO
in clientapi, be moved to userapi.

Forks clientapi/producers/syncapi.go to pushserver/ for later extension.

* Add clientapi routes for push rules to Pushserver.

A cleanup would be to move more of the name-splitting logic into
pushrules.go, to depollute routing.go.

* Output rooms.join.unread_notifications in /sync.

This is the read-side. Pushserver will be the write-side.

* Implement pushserver/storage for notifications.

* Use PushGatewayClient and the pushrules module in Pushserver's room consumer.

* Use one goroutine per user to avoid locking up the entire server for
  one bad push gateway.
* Split pushing by format.
* Send one device per push. Sytest does not support coalescing
  multiple devices into one push. Matches Synapse. Either we change
  Sytest, or remove the group-by-url-and-format logic.
* Write OutputNotificationData from push server. Sync API is already
  the consumer.

* Implement read receipt consumers in Pushserver.

Supports m.read and m.fully_read receipts.

* Add clientapi route for /unstable/notifications.

* Rename to UpsertPusher for clarity and handle pusher update

* Fix linter errors

* Ignore body.Close() error check

* Fix push server internal http wiring

* Add 40 newly passing 61push tests to whitelist

* Add next 12 newly passing 61push tests to whitelist

* Send notification data before notifying users in EDU server consumer

* NATS JetStream

* Goodbye sarama

* Fix `NewStreamTokenFromString`

* Consume on the correct topic for the roomserver

* Don't panic, NAK instead

* Move push notifications into the User API

* Don't set null values since that apparently causes Element upsetti

* Also set omitempty on conditions

* Fix bug so that we don't override the push rules unnecessarily

* Tweak defaults

* Update defaults

* More tweaks

* Move `/notifications` onto `r0`/`v3` mux

* User API will consume events and read/fully read markers from the sync API with stream positions, instead of consuming directly

Co-authored-by: Piotr Kozimor <p1996k@gmail.com>
Co-authored-by: Tommie Gannert <tommie@gannert.se>
Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-03-03 11:40:53 +00:00
Neil Alexander
726529fe99
Hopefully fix read receipts (#2241) 2022-03-01 16:59:11 +00:00
S7evinK
cf525d1f61
Implement /context (#2207)
* Add QueryEventsAfter

* Add /context

* Make all tests pass on sqlite

* Add queries to get the events for /context requests

* Move /context to the syncapi

* Revert "Add QueryEventsAfter"

This reverts commit 440a771d10632622e8c65d35fe90f0804bc98862.

* Simplify getting the required events

* Apply RoomEventFilter when getting events

* Add passing tests

* Remove logging

* Remove unused SQL statements
Update comments & add TODO
2022-02-21 17:12:22 +01:00
Neil Alexander
837f50ac89
Reduce CPU usage of SelectStateInRange (#2038) 2021-11-03 09:53:37 +00:00
Neil Alexander
f63068df3b
Only include go-sqlite3 on the relevant binaries (#1900)
* Only include go-sqlite3 on the relevant binaries

* The driver name is always sqlite3 now

* Update to matrix-org/go-sqlite3-js@e537baa
2021-07-20 11:18:14 +01:00
Neil Alexander
d15836e260
Increase gocyclo complexity to 25 (and remove all but 2 golint directives related to it) (#1783) 2021-03-03 14:35:57 +00:00
Neil Alexander
c9f305f254
Don't exclude an event from sync if it was previously not excluded (#1767) 2021-02-17 15:18:41 +00:00
Neil Alexander
6e44450cc9
Don't re-request state events that are already in the timeline (#1739)
* Don't request state events if we already have the timeline events (Postgres only)

* Rename variable

* nocyclo

* Add SQLite

* Tweaks

* Revert query change

* Don't dedupe if asking for full state

* Update query
2021-02-04 12:20:37 +00:00
Neil Alexander
b7e3b81a22
Fix ON CONFLICT on sync API account data (#1745) (#1750) 2021-02-04 11:45:49 +00:00