Previously, we relied on the plugin to close the DB connections
on shutdown. While this keeps the code simple, there is no guarantee
that the plugin author will remember to close the DB.
In that case, it's better to track the connections from the server side
and close them in case they weren't closed already. This complicates
the API slightly, but it's a price we need to pay.
https://mattermost.atlassian.net/browse/MM-56402
```release-note
We close any remaining unclosed DB RPC connections
after a plugin shuts down.
```
Co-authored-by: Jesse Hallam <jesse.hallam@gmail.com>
Co-authored-by: Mattermost Build <build@mattermost.com>
* always ping on plugin registration; SharedChannel.IsValid allow no team for GM
* wait for services to start before ping
* ping plugin remotes synchronously on startup
* remove the waitForInterClusterServices stuff
* don't set remoteid when inviting remote to channel
* Update server/public/model/remote_cluster_test.go
Co-authored-by: Ibrahim Serdar Acikgoz <serdaracikgoz86@gmail.com>
* address review comments
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
Co-authored-by: Ibrahim Serdar Acikgoz <serdaracikgoz86@gmail.com>
* Revert "Revert "MM-57759: Bump mockery to version 2.42.2 to support go 1.22^" (#26772)"
This reverts commit cd3b5b46e1.
* Added the hooks.go file changes as well
```release-note
NONE
```
* Allow end users to fetch the group members list of groups which allow @-mentions
* Update server/channels/api4/group_test.go
Co-authored-by: Ibrahim Serdar Acikgoz <serdaracikgoz86@gmail.com>
* Fix test name
* Move into subtest
---------
Co-authored-by: Ibrahim Serdar Acikgoz <serdaracikgoz86@gmail.com>
* ProfileImageBytes for EnsureBotOptions
* leverage plugintest.NewAPI
* fix linting
* add UpdateUserRoles to plugin api
* MM-57018: support reattaching plugins
Expose a local-only API for reattaching plugins: instead of the server starting and managing the process itself, allow the plugin to be launched externally (eg within a unit test) and reattach to an existing server instance to provide the unit test with a fully functional RPC API, sidestepping the need for mocking the plugin API in most cases.
In the future, this may become the basis for running plugins in a sidecar container.
Fixes: https://mattermost.atlassian.net/browse/MM-57018
* drop unused supervisor.pid
* factor out checkMinServerVersion
* factor out startPluginServer
* restore missing setPluginState on successful reattach
* avoid passing around a stale registeredPlugin
* inline initializePluginImplementation
* have IsValid return an error
* explicitly close rpcClient
In the case of reattached plugins, the Unix socket won't necessarily disappear leaving the muxBrokers blocked indefinitely. And `Kill()` doesn't do anything if there's no process being managed.
* explicitly detachPlugin
* emphasize gRPC not being supported
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
- Bump timeouts for some cases
- Improve sleep with require/assert.EventuallyT
for better performance and reliability.
https://mattermost.atlassian.net/browse/MM-57532
```release-note
NONE
```
Improvements to logging start-up.
- comments have been added
- order of configuration is adjusted to allow the main logger to capture errors in notification logger
- redundant logger initialization removed; no need to configure the logger just to replace the configuration immediately after
Previous log lines just had generic errors
which made things difficult to debug because
if a system bot is disabled, it's hard to know
whether the error is coming while trying to send
post reminders or from some other part of the system.
```release-note
NONE
```
* Added hard limits when creating user
* Added check to user activation
* Added missing check for licensed servers
* Fix i18n
* Fixed style order
* Added a separate hard limit along with existing 10k user soft limit
* For CI
* Fixing flaky test, hopefully
* Added tests
* add ratelimit to /sessions; cap userAgent version length; tests
* add MaxSessionsLimit; remove oldest session first; tests
* can't use slices in 1.20; improve test
* nits
* add GetLRUSessions; move limiting to CreateSession; remove rate limiting
* use queryBuilder
* mysql needs a limit when using offset
* update i18n
* refactor into limitNumberOfSessions; protect createSessionForUserAccessToken
* add comment to GetLRUSessions
* add limit to oauth path; PR comments
* Use a row constructor comparison to avoid a filter
The original query had a WHERE condition that looked like:
(x > a) OR (x = a AND y < b)
This commit changes it to a compound comparison that, given the
lexicographical order it uses, is semantically the same:
(x, y) > (a, b)
This makes the plan in Postgres use an Index Cond instead of a Filter,
having two main performance improvements:
1. The query's time is more or less constant, not increasing every time
this query is executed. This is important because the query is executed
repeatedly until all posts are indexed.
2. The query's time is much shorter: while the original query eventually
takes 30s, hitting the timeout, the new one takes ~30ms.
Notice the difference in the plans below:
- the original query uses a `Filter` step, while the new one uses an
`IndexCond` + `Filter` step.
- the original query has a shared hit count of 40557115 (109GiB), while
the new one has a shared hit count of 13951 (109MiBA). This does not
show exactly how much data is read from disk, given it's a nested loop
node, but it gives us an idea of the amount of data processed in the
server.
Original plan:
agnivaltdb=> EXPLAIN (ANALYZE, BUFFERS) SELECT Posts.*, Channels.TeamId FROM Posts LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE Posts.CreateAt > '1687424888405' OR (Posts.CreateAt = '1687424888405' AND Posts.Id > 'tpomh9yu1tffmdp6dopobwuc9h') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000;
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Limit (cost=1.22..2287.94 rows=10000 width=464) (actual time=23921.236..23954.229 rows=10000 loops=1)
Buffers: shared hit=40557115
-> Incremental Sort (cost=1.22..13998881.93 rows=61217938 width=464) (actual time=23921.235..23953.205 rows=10000 loops=1)
Sort Key: posts.createat, posts.id
Presorted Key: posts.createat
Full-sort Groups: 311 Sort Method: quicksort Average Memory: 45kB Peak Memory: 45kB
Buffers: shared hit=40557115
-> Nested Loop Left Join (cost=1.00..11421751.31 rows=61217938 width=464) (actual time=23920.970..23947.067 rows=10001 loops=1)
Buffers: shared hit=40557115
-> Index Scan using idx_posts_create_at on posts (cost=0.57..9889434.54 rows=61217938 width=461) (actual time=23920.930..23931.063 rows=10001 loops=1)
Filter: ((createat > '1687424888405'::bigint) OR ((createat = '1687424888405'::bigint) AND ((id)::text > 'tpomh9yu1tffmdp6dopobwuc9h'::text)))
Rows Removed by Filter: 40920000
Buffers: shared hit=40553119
-> Memoize (cost=0.43..0.70 rows=1 width=30) (actual time=0.001..0.001 rows=1 loops=10001)
Cache Key: posts.channelid
Cache Mode: logical
Hits: 9002 Misses: 999 Evictions: 0 Overflows: 0 Memory Usage: 151kB
Buffers: shared hit=3996
-> Index Scan using channels_pkey on channels (cost=0.42..0.69 rows=1 width=30) (actual time=0.007..0.007 rows=1 loops=999)
Index Cond: ((id)::text = (posts.channelid)::text)
Buffers: shared hit=3996
Planning:
Buffers: shared hit=112
Planning Time: 0.501 ms
Execution Time: 23954.974 ms
(25 rows)
New plan:
agnivaltdb=> EXPLAIN (ANALYZE, BUFFERS) SELECT Posts.*, Channels.TeamId FROM Posts LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE (Posts.CreateAt, Posts.Id) > ('1687424888405', 'tpomh9yu1tffmdp6dopobwuc9h') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000;
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
Limit (cost=1.15..1666.83 rows=10000 width=464) (actual time=0.366..34.053 rows=10000 loops=1)
Buffers: shared hit=13951
-> Incremental Sort (cost=1.15..10196977.36 rows=61217938 width=464) (actual time=0.365..33.031 rows=10000 loops=1)
Sort Key: posts.createat, posts.id
Presorted Key: posts.createat
Full-sort Groups: 311 Sort Method: quicksort Average Memory: 45kB Peak Memory: 45kB
Buffers: shared hit=13951
-> Nested Loop Left Join (cost=1.00..7619846.74 rows=61217938 width=464) (actual time=0.059..26.840 rows=10001 loops=1)
Buffers: shared hit=13951
-> Index Scan using idx_posts_create_at on posts (cost=0.57..6087529.97 rows=61217938 width=461) (actual time=0.040..10.548 rows=10001 loops=1)
Index Cond: (createat >= '1687424888405'::bigint)
Filter: (ROW(createat, (id)::text) > ROW('1687424888405'::bigint, 'tpomh9yu1tffmdp6dopobwuc9h'::text))
Rows Removed by Filter: 2
Buffers: shared hit=9955
-> Memoize (cost=0.43..0.70 rows=1 width=30) (actual time=0.001..0.001 rows=1 loops=10001)
Cache Key: posts.channelid
Cache Mode: logical
Hits: 9002 Misses: 999 Evictions: 0 Overflows: 0 Memory Usage: 151kB
Buffers: shared hit=3996
-> Index Scan using channels_pkey on channels (cost=0.42..0.69 rows=1 width=30) (actual time=0.007..0.007 rows=1 loops=999)
Index Cond: ((id)::text = (posts.channelid)::text)
Buffers: shared hit=3996
Planning:
Buffers: shared hit=112
Planning Time: 0.471 ms
Execution Time: 34.716 ms
(26 rows)
* Go back to the old query for MySQL
As one could expect, the two databases have completely opposite
behaviours, and for MySQL, the old query is significantly faster than
the new one, exactly for the same reason than in Postgres, but the other
way around: the old query uses an Index range scan while the new one
would use a Filter.
Old query (0.18s):
mysql> EXPLAIN ANALYZE SELECT Posts.*, Channels.TeamId FROM Posts USE INDEX(idx_posts_create_at_id) LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE Posts.CreateAt > 1557752415221 OR (Posts.CreateAt = 1557752415221 AND Posts.Id > 'ad59ire57tfwmjr5r8xqxc75qw') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000;
--------------
EXPLAIN ANALYZE SELECT Posts.*, Channels.TeamId FROM Posts USE INDEX(idx_posts_create_at_id) LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE Posts.CreateAt > 1557752415221 OR (Posts.CreateAt = 1557752415221 AND Posts.Id > 'ad59ire57tfwmjr5r8xqxc75qw') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000
--------------
+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| EXPLAIN |
+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| -> Limit: 10000 row(s) (cost=7472068.91 rows=10000) (actual time=0.063..164.174 rows=10000 loops=1)
-> Nested loop left join (cost=7472068.91 rows=5559093) (actual time=0.062..163.450 rows=10000 loops=1)
-> Index range scan on Posts using idx_posts_create_at_id over (CreateAt = 1557752415221 AND 'ad59ire57tfwmjr5r8xqxc75qw' < Id) OR (1557752415221 < CreateAt), with index condition: ((Posts.CreateAt > 1557752415221) or ((Posts.CreateAt = 1557752415221) and (Posts.Id > 'ad59ire57tfwmjr5r8xqxc75qw'))) (cost=1357066.29 rows=5559093) (actual time=0.043..97.358 rows=10000 loops=1)
-> Single-row index lookup on Channels using PRIMARY (Id=Posts.ChannelId) (cost=1.00 rows=1) (actual time=0.006..0.006 rows=1 loops=10000)
|
+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.18 sec)
New query (5s):
mysql> EXPLAIN ANALYZE SELECT Posts.*, Channels.TeamId FROM Posts USE INDEX(idx_posts_create_at_id) LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE (Posts.CreateAt, Posts.Id) > (1557752415221, 'ad
59ire57tfwmjr5r8xqxc75qw') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000;
--------------
EXPLAIN ANALYZE SELECT Posts.*, Channels.TeamId FROM Posts USE INDEX(idx_posts_create_at_id) LEFT JOIN Channels ON Posts.ChannelId = Channels.Id WHERE (Posts.CreateAt, Posts.Id) > (1557752415221, 'ad59ire57tfwmjr5r8xqxc75qw') ORDER BY Posts.CreateAt ASC, Posts.Id ASC LIMIT 10000
--------------
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| EXPLAIN |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| -> Limit: 10000 row(s) (cost=11119405.48 rows=10000) (actual time=5295.106..5455.285 rows=10000 loops=1)
-> Nested loop left join (cost=11119405.48 rows=10000) (actual time=5295.105..5454.572 rows=10000 loops=1)
-> Filter: ((Posts.CreateAt,Posts.Id) > (1557752415221,'ad59ire57tfwmjr5r8xqxc75qw')) (cost=221.48 rows=10000) (actual time=5295.078..5388.668 rows=10000 loops=1)
-> Index scan on Posts using idx_posts_create_at_id (cost=221.48 rows=10000) (actual time=0.055..5314.753 rows=600000 loops=1)
-> Single-row index lookup on Channels using PRIMARY (Id=Posts.ChannelId) (cost=1.00 rows=1) (actual time=0.006..0.006 rows=1 loops=10000)
|
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (5.46 sec)
```
* Add comment explaining the difference between DBs
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
We use the same name for hit/miss/invalidation metrics.
The downside is that now we have conflated a full cache purge
and a single cache key invalidation into a single label, but
the advantage is that it's easier to monitor the usage of a single
cache.
And moreover, duplicating the same string across multiple places
increases the chances of making a mistake. Centralizing the cache
name only at instantiation simplifies the code.
After this, we would be ready to use Redis.
https://mattermost.atlassian.net/browse/MM-57193
```release-note
NONE
```
