Commit Graph

274 Commits

Author SHA1 Message Date
Neil Alexander
131bedc1a1
Remove event type and state key caches (#2200)
* Don't proactively cache event types and state keys when we don't know if the transaction has persisted yet

* Remove event type and state key caches altogether
2022-02-18 10:58:41 +00:00
Neil Alexander
0b123b29f5
Use process context for roomserver input (#2198) 2022-02-17 15:58:54 +00:00
Neil Alexander
7dfc7c3d70
Don't re-send sent events in add_state_events (#2195)
* Only add events to `add_state_events` that haven't already been sent to the roomserver output before

* Filter on event NIDs instead, hopefully bring joy to SQLite

* UnsentFilter, review comments
2022-02-17 13:53:48 +00:00
Neil Alexander
5106cc807c
Ensure only one transaction is used for RS input per room (#2178)
* Ensure the input API only uses a single transaction

* Remove more of the dead query API call

* Tidy up

* Fix tests hopefully

* Don't do unnecessary work for rooms that don't exist

* Improve error, fix another case where transaction wasn't used properly

* Add a unit test for checking single transaction on RS input API

* Fix logic oops when deciding whether to use a transaction in storeEvent
2022-02-11 17:40:14 +00:00
Neil Alexander
2782ae3d56
Fix fetching missing state (#2163)
* Check that we have a populated state snapshot when determining if we closed the gap

* Do the same in the query API

* Use HasState more opportunistically

* Try to avoid falling down the hole of using a trustworthy but empty state snapshot for non-create events

* Refactor missing state and make sure that we really solve the problem for the new event

* Comments

* Review comments

* Tweak that check again

* Tidy up that create check further

* Fix build hopefully

* Update sendOutliers to use OrderAuthAndStateEvents

* Don't go out of bounds on missingEvents
2022-02-10 10:05:14 +00:00
Neil Alexander
37cbe263ce
Fix transaction issues in events table in PSQL (#2165)
* Revert "Revert "Fix storage bug in PSQL events table""

This reverts commit cf447dd52a.

* Membership updater to use updater

* Fix membership updater to use transactions properly
2022-02-10 09:30:16 +00:00
kegsay
aa5c3b88de
Unmarshal events at the Dendrite level not GMSL level (#2164)
* Use new event json types in gmsl

* Fix EventJSON to actually unmarshal events

* Update GMSL

* Bump GMSL and improve error messages

* Send back the correct RespState

* Update GMSL
2022-02-09 20:31:24 +00:00
Neil Alexander
cf447dd52a
Revert "Fix storage bug in PSQL events table"
This reverts commit b4687f2ed2.
2022-02-09 11:41:21 +00:00
Neil Alexander
b4687f2ed2
Fix storage bug in PSQL events table 2022-02-09 11:24:49 +00:00
Neil Alexander
457a07eac5
More relaxed auth event fetching (#2161)
* Tweaks around auth event fetching

* More tweaking
2022-02-08 17:06:13 +00:00
Neil Alexander
0e26662a55
Allow events to be un-rejected (#2159)
* Allow un-rejecting an event later

* SQL

* Only un-reject, don't re-reject

* Clarify ambiguous column reference
2022-02-08 13:45:48 +00:00
Neil Alexander
a572f4db03
Fix bugs that could wedge rooms (#2154)
* Don't flake so badly for rejected events

* Moar

* Fix panic

* Don't count rejected events as missing

* Don't treat rejected events without state as missing

* Revert "Don't count rejected events as missing"

This reverts commit 4b6139b62eb91ba059b47415b0275964b37d9b43.

* Missing events should be KindOld

* If we have state, use it, regardless of memberships which could be stale now

* Fetch missing state for KindOld too

* Tweak the condition again

* Clean up a bit

* Use room updater to get latest events in a race-free way

* Return the correct error

* Improve errors
2022-02-07 19:10:01 +00:00
S7evinK
9de7efa0b0
Remove sarama/saramajetstream dependencies (#2138)
* Remove dependency on saramajetstream & sarama

Signed-off-by: Till Faelligen <tfaelligen@gmail.com>

* Remove internal.ContinualConsumer from federationapi

* Remove internal.ContinualConsumer from syncapi

* Remove internal.ContinualConsumer from keyserver

* Move to new Prepare function

* Remove saramajetstream & sarama dependency

* Delete unneeded file

* Remove duplicate import

* Log error instead of silently irgnoring it

* Move `OffsetNewest` and `OffsetOldest` into keyserver types, change them to be more sane values

* Fix comments

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-02-04 13:08:13 +00:00
Neil Alexander
532f445c4e
Remove roomserver input deadlines (#2144)
It isn't really clear that the deadlines actually help in any way. Currently we can use up our 2 minutes doing something, run out of context time and then return an error which causes the transaction to rollback and forgetting everything we've done. If the message came to us from NATS then we probably will end up retrying just to be in the same situation. We'd be really a lot better if we just spent the time reconciling the problem in the first place, and then we're much less likely to need to fetch those missing auth or prev events in the future.

Also includes matrix-org/gomatrixserverlib#287 so we don't wait so long for servers that are obviously dead.
2022-02-04 12:13:07 +00:00
Neil Alexander
eb352a5f6b
Full roomserver input transactional isolation (#2141)
* Add transaction to all database tables in roomserver, rename latest events updater to room updater, use room updater for all RS input

* Better transaction management

* Tweak order

* Handle cases where the room does not exist

* Other fixes

* More tweaks

* Fill some gaps

* Fill in the gaps

* good lord it gets worse

* Don't roll back transactions when events rejected

* Pass through errors properly

* Fix bugs

* Fix incorrect error check

* Don't panic on nil txns

* Tweaks

* Hopefully fix panics for good in SQLite this time

* Fix rollback

* Minor bug fixes with latest event updater

* Some review comments

* Revert "Some review comments"

This reverts commit 0caf8cf53e62c33f7b83c52e9df1d963871f751e.

* Fix a couple of bugs

* Clearer commit and rollback results

* Remove unnecessary prepares
2022-02-04 10:39:34 +00:00
Neil Alexander
4d9f5b2e57
Fix panic from closing the input channel before the workers complete (it'll get GC'd either way) 2022-02-02 17:46:37 +00:00
Neil Alexander
c773b038bb
Use pull consumers (#2140)
* Pull consumers

* Pull consumers

* Only nuke consumers if they are push consumers

* Clean up old consumers

* Better error handling

* Update comments
2022-02-02 13:32:48 +00:00
kegsay
2dee706f9e
PerformInvite: bugfix and rejig control flow (#2137)
* PerformInvite: bugfix and rejig control flow

Local clients would not be notified of invites to rooms
Dendrite had already joined in all cases due to not returning
an `api.OutputNewInviteEvent` for local invites. We now do this.

This was an easy mistake to make due to the control flow of the
function which doesn't handle the happy case at the end of the
function and instead forks the function depending on if the
invite was via federation or not. This has now been changed to
handle the federated invite as if it were an error (in that we
check it, do it and bail out) rather than outstay our welcome.
This ends up with the local invite being the happy case, which
now both sends an `InputRoomEvent` to the roomserver _and_ a
`api.OutputNewInviteEvent` is returned.

* Don't send invite pokes in PerformInvite

* Move event ID into logger
2022-02-02 13:30:48 +00:00
Neil Alexander
893aa3b141
More logging tweaks 2022-01-31 16:01:54 +00:00
Neil Alexander
07d0e72a8b
Improve roomserver logging 2022-01-31 15:33:00 +00:00
Neil Alexander
d21f3eace0
Roomserver fixes (#2133)
* Improve server selection somewhat

* Remove things from the map when we're done

* Be less panicky about auth event signatures in case they are not fatal after all

* Accept HasState in all cases

* Send join asynchronously

* Revert "Send join asynchronously"

This reverts commit 5b685bfcd0b1150a66c7b1e70fb3a3eda509efd1.

* Joins and leaves use background context
2022-01-31 14:36:59 +00:00
Neil Alexander
f9547a53d2
Tweak roomserver logging for rejected events 2022-01-31 12:01:53 +00:00
Neil Alexander
ba1a9b98b7
Tweak some logging (#2130)
* Modify some log levels

* Update gomatrixserverlib to matrix-org/gomatrixserverlib@336334f

* Update gomatrixserverlib to matrix-org/gomatrixserverlib@cde7ac8

* Demote warning about key change producer

* Add more useful roomserver logging

* Further tweaking
2022-01-31 10:48:28 +00:00
Neil Alexander
eb8e770e99
Revert consumer change 2022-01-31 10:42:41 +00:00
Neil Alexander
a271fde8f5
Only limit context for fetching missing auth/prev events (#2131) 2022-01-31 10:39:33 +00:00
Neil Alexander
8e4002831f
Call hooks for outliers (#2119)
* Move hook call when processing room events

* Fix build

* Call hooks for outliers too
2022-01-28 13:11:56 +00:00
Neil Alexander
e9fbad6f20
Move hook call when processing room events (#2118)
* Move hook call when processing room events

* Fix build
2022-01-28 12:33:31 +00:00
Neil Alexander
48789ebec5
Don't flood Sentry with context cancelled/deadline exceeded errors (#2115) 2022-01-28 10:27:28 +00:00
Neil Alexander
b6011d0d87
Try federation when backfill fails to find events in the database (#2113)
* Try to backfill via federation in error cases

* Cleaner retry for backfill

* Simpler condition
2022-01-27 15:52:32 +00:00
Neil Alexander
a763cbb0e1
Roomserver/federation input refactor (#2104)
* Put federation client functions into their own file

* Look for missing auth events in RS input

* Remove retrieveMissingAuthEvents from federation API

* Logging

* Sorta transplanted the code over

* Use event origin failing all else

* Don't get stuck on mutexes:

* Add verifier

* Don't mark state events with zero snapshot NID as not existing

* Check missing state if not an outlier before storing the event

* Reject instead of soft-fail, don't copy roominfo so much

* Use synchronous contexts, limit time to fetch missing events

* Clean up some commented out bits

* Simplify `/send` endpoint significantly

* Submit async

* Report errors on sending to RS input

* Set max payload in NATS to 16MB

* Tweak metrics

* Add `workerForRoom` for tidiness

* Try skipping unmarshalling errors for RespMissingEvents

* Track missing prev events separately to avoid calculating state when not possible

* Tweak logic around checking missing state

* Care about state when checking missing prev events

* Don't check missing state for create events

* Try that again

* Handle create events better

* Send create room events as new

* Use given event kind when sending auth/state events

* Revert "Use given event kind when sending auth/state events"

This reverts commit 089d64d271b5fca8c104e1554711187420dbebca.

* Only search for missing prev events or state for new events

* Tweaks

* We only have missing prev if we don't supply state

* Room version tweaks

* Allow async inputs again

* Apply backpressure to consumers/synchronous requests to hopefully stop things being overwhelmed

* Set timeouts on roomserver input tasks (need to decide what timeout makes sense)

* Use work queue policy, deliver all on restart

* Reduce chance of duplicates being sent by NATS

* Limit the number of servers we attempt to reduce backpressure

* Some review comment fixes

* Tidy up a couple things

* Don't limit servers, randomise order using map

* Some context refactoring

* Update gmsl

* Don't resend create events

* Set stateIDs length correctly or else the roomserver thinks there are missing events when there aren't

* Exclude our own servername

* Try backing off servers

* Make excluding self behaviour optional

* Exclude self from g_m_e

* Update sytest-whitelist

* Update consumers for the roomserver output stream

* Remember to send outliers for state returned from /gme

* Make full HTTP tests less upsetti

* Remove 'If a device list update goes missing, the server resyncs on the next one' from the sytest blacklist

* Remove debugging test

* Fix blacklist again, remove unnecessary duplicate context

* Clearer contexts, don't use background in case there's something happening there

* Don't queue up events more than once in memory

* Correctly identify create events when checking for state

* Fill in gaps again in /gme code

* Remove `AuthEventIDs` from `InputRoomEvent`

* Remove stray field

Co-authored-by: Kegan Dougal <kegan@matrix.org>
2022-01-27 14:29:14 +00:00
kegsay
0bf5104bbb
Fix #2027 by gracefully handling stub rooms (#2100)
The server ACL code on startup will grab all known rooms from
the rooms_table and then call `GetStateEvent` with each found
room ID to find the server ACL event. This can fail for stub
rooms, which will be present in the rooms table. Previously
this would result in an error being returned and the server
failing to start (!). Now we just return no event for stub
rooms.
2022-01-21 14:23:37 +00:00
Neil Alexander
16035b9737
NATS JetStream tweaks (#2086)
* Use named NATS durable consumers

* Build fixes

* Remove dupe call to SetFederationAPI

* Use namespaced consumer name

* Fix namespacing

* Fix unit tests hopefully
2022-01-07 17:31:57 +00:00
Neil Alexander
a422321435
Fix panic at startup if roomserver was not given federation API reference by the time NATS consumes an event, tweak backpressure metrics 2022-01-07 13:41:53 +00:00
kegsay
173b1e8d3e
Fix #2084 - incorrect /event_auth response (#2085)
* Fix #2084

* Return early

* Linting
2022-01-06 17:13:34 +00:00
S7evinK
161f145176
Add NATS JetStream support (#1866)
* Add NATS JetStream support
Update shopify/sarama

* Fix addresses

* Don't change Addresses in Defaults

* Update saramajetstream

* Add missing error check

Keep typing events for at least one minute

* Use all configured NATS addresses

* Update saramajetstream

* Try setting up with NATS

* Make sure NATS uses own persistent directory (TODO: make this configurable)

* Update go.mod/go.sum

* Jetstream package

* Various other refactoring

* Build fixes

* Config tweaks, make random jetstream storage path for CI

* Disable interest policies

* Try to sane default on jetstream base path

* Try to use in-memory for CI

* Restore storage/retention

* Update nats.go dependency

* Adapt changes to config

* Remove unneeded TopicFor

* Dep update

* Revert "Remove unneeded TopicFor"

This reverts commit f5a4e4a339b6f94ec215778dca22204adaa893d1.

* Revert changes made to streams

* Fix build problems

* Update nats-server

* Update go.mod/go.sum

* Roomserver input API queuing using NATS

* Fix topic naming

* Prometheus metrics

* More refactoring to remove saramajetstream

* Add missing topic

* Don't try to populate map that doesn't exist

* Roomserver output topic

* Update go.mod/go.sum

* Message acknowledgements

* Ack tweaks

* Try to resume transaction re-sends

* Try to resume transaction re-sends

* Update to matrix-org/gomatrixserverlib@91dadfb

* Remove internal.PartitionStorer from components that don't consume keychanges

* Try to reduce re-allocations a bit in resolveConflictsV2

* Tweak delivery options on RS input

* Publish send-to-device messages into correct JetStream subject

* Async and sync roomserver input

* Update dendrite-config.yaml

* Remove roomserver tests for now (they need rewriting)

* Remove roomserver test again (was merged back in)

* Update documentation

* Docker updates

* More Docker updates

* Update Docker readme again

* Fix lint issues

* Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset)

* Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that

* Go 1.16 instead of Go 1.13 for upgrade tests and Complement

* Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that"

This reverts commit 368675283fc44501f227639811bdb16dd5deef8c.

* Don't report any errors on `/send` to see what fun that creates

* Fix panics on closed channel sends

* Enforce state key matches sender

* Do the same for leave

* Various tweaks to make tests happier

Squashed commit of the following:

commit 13f9028e7a63662759ce7c55504a9d2423058668
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 15:47:14 2022 +0000

    Do the same for leave

commit e6be7f05c349fafbdddfe818337a17a60c867be1
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 15:33:42 2022 +0000

    Enforce state key matches sender

commit 85ede6d64bf10ce9b91cdd6d80f87350ee55242f
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 14:07:04 2022 +0000

    Fix panics on closed channel sends

commit 9755494a98bed62450f8001d8128e40481d27e15
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 13:38:22 2022 +0000

    Don't report any errors on `/send` to see what fun that creates

commit 3bb4f87b5dd56882febb4db5621db484c8789b7c
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 13:00:26 2022 +0000

    Revert "Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that"

    This reverts commit 368675283fc44501f227639811bdb16dd5deef8c.

commit fe2673ed7be9559eaca134424e403a4faca100b0
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 12:09:34 2022 +0000

    Go 1.16 instead of Go 1.13 for upgrade tests and Complement

commit 368675283fc44501f227639811bdb16dd5deef8c
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 11:51:45 2022 +0000

    Don't report event rejection errors via `/send`, since apparently this is upsetting tests that don't expect that

commit b028dfc08577bcf52e6cb498026e15fa5d46d07c
Author: Neil Alexander <neilalexander@users.noreply.github.com>
Date:   Tue Jan 4 10:29:08 2022 +0000

    Send final event in `processEvent` synchronously (since this might stop Sytest from being so upset)

* Merge in NATS Server v2.6.6 and nats.go v1.13 into the in-process connection fork

* Add `jetstream.WithJetStreamMessage` to make ack/nak-ing less messy, use process context in consumers

* Fix consumer component name in  federation API

* Add comment explaining where streams are defined

* Tweaks to roomserver input with comments

* Finish that sentence that I apparently forgot to finish in INSTALL.md

* Bump version number of config to 2

* Add comments around asynchronous sends to roomserver in processEventWithMissingState

* More useful error message when the config version does not match

* Set version in generate-config

* Fix version in config.Defaults

Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2022-01-05 17:44:49 +00:00
Neil Alexander
3113210f17
Fix keyring regressions in previous P2P demo 2021-12-13 13:24:49 +00:00
Neil Alexander
c3dda0779d
Return event NID from StoreEvent, match PSQL vs SQLite behaviour, tweak backfill persistence (#2071) 2021-12-09 15:03:26 +00:00
Neil Alexander
c9419e51af
Don't populate config defaults where it doesn't make sense (#2058)
* Don't populate config defaults where it doesn't make sense

* Fix dendritejs builds
2021-11-24 11:57:39 +00:00
Neil Alexander
ec716793eb
Merge federationapi, federationsender, signingkeyserver components (#2055)
* Initial federation sender -> federation API refactoring

* Move base into own package, avoids import cycle

* Fix build errors

* Fix tests

* Add signing key server tables

* Try to fold signing key server into federation API

* Fix dendritejs builds

* Update embedded interfaces

* Fix panic, fix lint error

* Update configs, docker

* Rename some things

* Reuse same keyring on the implementing side

* Fix federation tests, `NewBaseDendrite` can accept freeform options

* Fix build

* Update create_db, configs

* Name tables back

* Don't rename federationsender consumer for now
2021-11-24 10:45:23 +00:00
Neil Alexander
6e93531e94
Don't persist transaction IDs in the roomserver (#2048) 2021-11-22 09:13:12 +00:00
Neil Alexander
403498a85b
Only return non-stub rooms from GetKnownRooms (#2049)
* Only return non-stub rooms from `GetKnownRooms`

This should stop a bunch of errors at startup with invalid server ACLs.

* Fix query
2021-11-18 11:34:19 +00:00
Neil Alexander
b9a575919a
Try to reduce re-allocations a bit in resolveConflictsV2 2021-11-04 10:55:07 +00:00
PiotrKozimor
dec05c3347
Run gofmt on dendrite - apply go 1.17 preferred build tags (#2021) 2021-11-02 16:48:48 +00:00
Neil Alexander
b99f594a93
Fix #2028 (#2036) 2021-11-02 16:47:39 +00:00
Ryan W
a624eab309
- Removed double imports (#1989)
- Lower cased error messages

Signed-off-by: Ryan Whittington <twentybitdev@gmail.com>

Co-authored-by: kegsay <kegan@matrix.org>
2021-09-08 17:31:03 +01:00
kegsay
7dc8fb1fe7
Add more logs (#2005)
* Add more logs

To help debug the migration issue in #1924 along with manual data-loss-inducing fixes.
Also log the origin server on processed txns to help debug buggy server origins.

* Fix query
2021-09-07 15:07:14 +01:00
Neil Alexander
51b119107c
Don't return nonsense canonical room aliases in the public rooms responses (#1992) 2021-08-27 16:50:30 +01:00
kegsay
4cc8b28b7f
Ensure all create events have a snapshot NID of 0 (#1961)
Fixes #1924 for postgres users, though the underlying cause of why
they aren't 0 in the first place is unresolved.
2021-08-04 17:48:23 +01:00
kegsay
ed04eed441
Fix sqlite migration issues (#1960)
* Do not store 'null' in the database for empty JSON arrays

This can cause issues, though it should be noted that the majority
of the time this will marshal/unmarshal just fine, see
https://play.golang.org/p/Doe2NZUgv7Q

* bugfix: sqlite migration should handle create events as having no 'before' snapshot

The state snapshot for any given event in the roomserver represents the state _before_
the event. For the create event, this is nothing, so the state snapshot nid should be 0.

In some cases this wasn't happening, resulting in a nice mix of possible options including:
 - A state snapshot without any state blocks `[]` or `null`.
 - A state snapshot with a single state block with a single event, the create event, causing
   a circular loop. This is incorrect as it represents the state before the event, not after.

* Add state key check
2021-08-04 17:08:17 +01:00
Kegan Dougal
ed4097825b Factor out StatementList to sqlutil and use it in userapi
It helps with the boilerplate.
2021-07-28 18:30:04 +01:00