12649ccedd
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>`
|
||
|---|---|---|
| .github | ||
| appservice | ||
| build | ||
| clientapi | ||
| cmd | ||
| docs | ||
| federationapi | ||
| internal | ||
| keyserver | ||
| mediaapi | ||
| roomserver | ||
| setup | ||
| syncapi | ||
| test | ||
| userapi | ||
| .dockerignore | ||
| .gitignore | ||
| .golangci.yml | ||
| are-we-synapse-yet.list | ||
| are-we-synapse-yet.py | ||
| build-dendritejs.sh | ||
| build.cmd | ||
| build.sh | ||
| CHANGES.md | ||
| dendrite-sample.monolith.yaml | ||
| dendrite-sample.polylith.yaml | ||
| go.mod | ||
| go.sum | ||
| LICENSE | ||
| README.md | ||
| run-sytest.sh | ||
| show-expected-fail-tests.sh | ||
| sytest-blacklist | ||
| sytest-whitelist | ||
| test-dendritejs.sh | ||
Dendrite
Dendrite is a second-generation Matrix homeserver written in Go. It intends to provide an efficient, reliable and scalable alternative to Synapse:
- Efficient: A small memory footprint with better baseline performance than an out-of-the-box Synapse.
- Reliable: Implements the Matrix specification as written, using the same test suite as Synapse as well as a brand new Go test suite.
- Scalable: can run on multiple machines and eventually scale to massive homeserver deployments.
Dendrite is beta software, which means:
- Dendrite is ready for early adopters. We recommend running in Monolith mode with a PostgreSQL database.
- Dendrite has periodic releases. We intend to release new versions as we fix bugs and land significant features.
- Dendrite supports database schema upgrades between releases. This means you should never lose your messages when upgrading Dendrite.
This does not mean:
- Dendrite is bug-free. It has not yet been battle-tested in the real world and so will be error prone initially.
- Dendrite is feature-complete. There may be client or federation APIs that are not implemented.
- Dendrite is ready for massive homeserver deployments. There is no sharding of microservices (although it is possible to run them on separate machines) and there is no high-availability/clustering support.
Currently, we expect Dendrite to function well for small (10s/100s of users) homeserver deployments as well as P2P Matrix nodes in-browser or on mobile devices.
In the future, we will be able to scale up to gigantic servers (equivalent to matrix.org) via polylith mode.
If you have further questions, please take a look at our FAQ or join us in:
- #dendrite:matrix.org - General chat about the Dendrite project, for users and server admins alike
- #dendrite-dev:matrix.org - The place for developers, where all Dendrite development discussion happens
- #dendrite-alerts:matrix.org - Release notifications and important info, highly recommended for all Dendrite server admins
Requirements
See the Planning your Installation page for more information on requirements.
To build Dendrite, you will need Go 1.18 or later.
For a usable federating Dendrite deployment, you will also need:
- A domain name (or subdomain)
- A valid TLS certificate issued by a trusted authority for that domain
- SRV records or a well-known file pointing to your deployment
Also recommended are:
- A PostgreSQL database engine, which will perform better than SQLite with many users and/or larger rooms
- A reverse proxy server, such as nginx, configured like this sample
The Federation Tester can be used to verify your deployment.
Get started
If you wish to build a fully-federating Dendrite instance, see the Installation documentation. For running in Docker, see build/docker.
The following instructions are enough to get Dendrite started as a non-federating test deployment using self-signed certificates and SQLite databases:
$ git clone https://github.com/matrix-org/dendrite
$ cd dendrite
$ ./build.sh
# Generate a Matrix signing key for federation (required)
$ ./bin/generate-keys --private-key matrix_key.pem
# Generate a self-signed certificate (optional, but a valid TLS certificate is normally
# needed for Matrix federation/clients to work properly!)
$ ./bin/generate-keys --tls-cert server.crt --tls-key server.key
# Copy and modify the config file - you'll need to set a server name and paths to the keys
# at the very least, along with setting up the database connection strings.
$ cp dendrite-sample.monolith.yaml dendrite.yaml
# Build and run the server:
$ ./bin/dendrite-monolith-server --tls-cert server.crt --tls-key server.key --config dendrite.yaml
# Create an user account (add -admin for an admin user).
# Specify the localpart only, e.g. 'alice' for '@alice:domain.com'
$ ./bin/create-account --config dendrite.yaml --url http://localhost:8008 --username alice
Then point your favourite Matrix client at http://localhost:8008 or https://localhost:8448.
Progress
We use a script called Are We Synapse Yet which checks Sytest compliance rates. Sytest is a black-box homeserver test rig with around 900 tests. The script works out how many of these tests are passing on Dendrite and it updates with CI. As of August 2022 we're at around 90% CS API coverage and 95% Federation coverage, though check CI for the latest numbers. In practice, this means you can communicate locally and via federation with Synapse servers such as matrix.org reasonably well, although there are still some missing features (like Search).
We are prioritising features that will benefit single-user homeservers first (e.g Receipts, E2E) rather than features that massive deployments may be interested in (OpenID, Guests, Admin APIs, AS API). This means Dendrite supports amongst others:
- Core room functionality (creating rooms, invites, auth rules)
- Room versions 1 to 10 supported
- Backfilling locally and via federation
- Accounts, profiles and devices
- Published room lists
- Typing
- Media APIs
- Redaction
- Tagging
- Context
- E2E keys and device lists
- Receipts
- Push
- Guests
- User Directory
- Presence
Contributing
We would be grateful for any help on issues marked as Are We Synapse Yet. These issues all have related Sytests which need to pass in order for the issue to be closed. Once you've written your code, you can quickly run Sytest to ensure that the test names are now passing.
If you're new to the project, see our Contributing page to get up to speed, then look for Good First Issues. If you're familiar with the project, look for Help Wanted issues.