* Correctly redact events over federation (#2526) * Ensure we check powerlevel/origin before redacting an event * Add passing test * Use pl.UserLevel * Make check more readable, also check for the sender * Add new next steps page to the documentation * Highlighting in docs * Rename the page to "Optimise your installation" * Attempt to raise the file descriptor limit at startup (#2527) * Add `--difference` to `resolve-state` tool * Make the linter happy again * generic CaddyFile in front of Dendrite (monolith) (#2531) for Caddy 2.5.x Co-authored-by: emanuele.aliberti <emanuele.aliberti@mtka.eu> * Handle state before, send history visibility in output (#2532) * Check state before event * Tweaks * Refactor a bit, include in output events * Don't waste time if soft failed either * Tweak control flow, comments, use GMSL history visibility type * Fix rare panic when returning user devices over federation (#2534) * Add `InputDeviceListUpdate` to the keyserver, remove old input API (#2536) * Add `InputDeviceListUpdate` to the keyserver, remove old input API * Fix copyright * Log more information when a device list update fails * Fix nats.go commit (#2540) Signed-off-by: Jean Lucas <jean@4ray.co> * Don't return `end` if there are not more messages (#2542) * Be more spec compliant * Move lazyLoadMembers to own method * Return an error if trying to invite a malformed user ID (#2543) * Add `evacuateUser` endpoint, use it when deactivating accounts (#2545) * Add `evacuateUser` endpoint, use it when deactivating accounts * Populate the API * Clean up user devices when deactivating * Include invites, delete pushers * Silence presence logs (#2547) * Blacklist `Guest users can join guest_access rooms` test until it can be investigated * Disable WebAssembly builds for now * Try to fix backfilling (#2548) * Try to fix backfilling * Return start/end to not confuse clients * Update GMSL * Update GMSL * Roomserver producers package (#2546) * Give the roomserver a producers package * Change init point * Populate ACLs API * Fix build issues * `RoomEventProducer` naming * Version 0.8.9 (#2549) * Version 0.8.9 * Update changelog * feat+fix: Ignore unknown keys and verify required fields are present in appservice registration files (#2550) * fix: ignore unknown keys in appservice configs fixes matrix-org/dendrite#1567 * feat: verify required fields in appservice configs * Use new testrig for key changes tests (#2552) * Use new testrig for tests * Log the error message * Fix QuerySharedUsers for the SyncAPI keychange consumer (#2554) * Make more use of base.BaseDendrite * Fix QuerySharedUsers if no UserIDs are supplied * Return clearer error when no state NID exists for an event (#2555) * Wrap error from `SnapshotNIDFromEventID` * Hopefully fix read receipts timestamps (#2557) This should avoid coercions between signed and unsigned ints which might fix problems like `sql: converting argument $5 type: uint64 values with high bit set are not supported`. * Fix nil pointer access when redacting events (#2560) * Fix issue `uint64 values with high bit are not supported` in presence (#2562) * Fix issue #2528 * Use gomatrixserverlib.Timestamp * Use ParseUint instead of ParseInt * Update Pinecone to matrix-org/pinecone@1ce778f * Ristretto cache (#2563) * Try Ristretto cache * Tweak * It's beautiful * Update GMSL * More strict keyable interface * Fix that some more * Make less panicky * Don't enforce mutability checks for now * Determine mutability using deep equality * Tweaks * Namespace keys * Make federation caches mutable * Update cost estimation, add metric * Update GMSL * Estimate cost for metrics better * Reduce counters a bit * Try caching events * Some guards * Try again * Try this * Use separate caches for hopefully better hash distribution * Fix bug with admitting events into cache * Try to fix bugs * Check nil * Try that again * Preserve order jeezo this is messy * thanks VS Code for doing exactly the wrong thing * Try this again * Be more specific * aaaaargh * One more time * That might be better * Stronger sorting * Cache expiries, async publishing of EDUs * Put it back * Use a shared cache again * Cost estimation fixes * Update ristretto * Reduce counters a bit * Clean up a bit * Update GMSL * 1GB * Configurable cache sizees * Tweaks * Add `config.DataUnit` for specifying friendly cache sizes * Various tweaks * Update GMSL * Add back some lazy loading caching * Include key in cost * Include key in cost * Tweak max age handling, config key name * Only register prometheus metrics if requested * Review comments @S7evinK * Don't return errors when creating caches (it is better just to crash since otherwise we'll `nil`-pointer exception everywhere) * Review comments * Update sample configs * Update GHA Workflow * Update Complement images to Go 1.18 * Remove the cache test from the federation API as we no longer guarantee immediate cache admission * Don't check the caches in the renewal test * Possibly fix the upgrade tests * Update to matrix-org/gomatrixserverlib#322 * Update documentation to refer to Go 1.18 * Minor SendToDevice fix (#2565) * Avoid unnecessary marshalling if sending to the local server * Fix ordering of ToDevice messages * Revive SendToDevice test * Use `/v3` to request media from remote servers (update to matrix-org/gomatrixserverlib#324) * Pointerise `types.RoomInfo` in the cache so we can update it in-place in the latest events updater * Add a Troubleshooting page * Update `sytest-whitelist` * 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` * Update 1_createusers.md (#2571) * Update 1_createusers.md Added description on how to create user accounts when running in docker. * Update 1_createusers.md Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com> * Fix connection_string format in dendrite-sample.polylith.yaml (#2574) * 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> * Tweak cache counters (#2575) * Tweak cache counters This makes the number of counters relative to the maximum cache size. Since the counters effectively manage the size of the bloom filter, larger caches need more counters and smaller caches need less. 10 counters per 1KB data means that the default cache size of 1GB should result in a bloom filter and TinyLRU admission set of about 16MB estimated. * Remove line left by accident * Set historyVisibility in rowsToStreamEvents * Update FAQ * Add event state key cache (#2576) * Explain how SRV works in Matrix and discourage using it (#2577) * Explain how SRV works in Matrix and discourage using it * Minor tweaks to formatting Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com> * Fix issue with membership event_nid being 0 (#2580) * docs: Add build page; correct proxy info; fix Caddy example (#2579) * Add build page; correct proxy info; fix Caddy example * Improve Caddyfile example * Apply review comments; add polylith Caddyfile * Bump tzinfo from 1.2.9 to 1.2.10 in /docs (#2584) Bumps [tzinfo](https://github.com/tzinfo/tzinfo) from 1.2.9 to 1.2.10. - [Release notes](https://github.com/tzinfo/tzinfo/releases) - [Changelog](https://github.com/tzinfo/tzinfo/blob/master/CHANGES.md) - [Commits](https://github.com/tzinfo/tzinfo/compare/v1.2.9...v1.2.10) --- updated-dependencies: - dependency-name: tzinfo dependency-type: indirect ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Membership updater refactoring (#2541) * Membership updater refactoring * Pass in membership state * Use membership check rather than referring to state directly * Delete irrelevant membership states * We don't need the leave event after all * Tweaks * Put a log entry in that I might stand a chance of finding * Be less panicky * Tweak invite handling * Don't freak if we can't find the event NID * Use event NID from `types.Event` * Clean up * Better invite handling * Placate the almighty linter * Blacklist a Sytest which is otherwise fine under Complement for reasons I don't understand * Fix the sytest after all (thanks @S7evinK for the spot) * Try to fix HTTP 500s on `/members` (#2581) * 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> * Add .well-known/matrix/client to clientapi (#2551) Signed-off-by: Jonathan Bartlett <jonathan@jonnobrow.co.uk> Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com> * Remove `room_id` field from MSC2946 stripped events (closes #2588) * Remove `goose` from Dockerfiles * Make the User API responsible for sending account data output events (#2592) * Make the User API responsible for sending account data output events * Clean up producer * Review comments * Update NATS Server and nats.go to use upstream * Set CORS headers for HTTP 404 and 405 errors (#2599) * Set CORS headers for the 404s * Use custom handlers, plus one for HTTP 405 too * Tweak setup * Add to muxes too * Tidy up some more * Use built-in HTTP 404 handler * Don't bother setting it for federation-facing * Optimise checking other servers allowed to see events (#2596) * Try optimising checking if server is allowed to see event * Fix error * Handle case where snapshot NID is 0 * Fix query * Update SQL * Clean up `CheckServerAllowedToSeeEvent` * Not supported on SQLite * Maybe placate the unit tests * Review comments * De-race `types.RoomInfo` (#2600) * De-race `CompleteSync` (#2601) The `err` was coming from outside of the goroutine and being written to by concurrent goroutines. * Version 0.9.0 (#2602) Co-authored-by: Till <2353100+S7evinK@users.noreply.github.com> Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com> Co-authored-by: Till Faelligen <davidf@element.io> Co-authored-by: Emanuele Aliberti <dev@mtka.eu> Co-authored-by: emanuele.aliberti <emanuele.aliberti@mtka.eu> Co-authored-by: Jean Lucas <jean@4ray.co> Co-authored-by: Kabir Kwatra <kabir@kwatra.me> Co-authored-by: andreever <52261463+andreever@users.noreply.github.com> Co-authored-by: Maximilian Gaedig <38767445+MaximilianGaedig@users.noreply.github.com> Co-authored-by: Tulir Asokan <tulir@maunium.net> Co-authored-by: Matt Holt <mholt@users.noreply.github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: kegsay <kegan@matrix.org> Co-authored-by: Jonathan Bartlett <34320158+Jonnobrow@users.noreply.github.com>
5.7 KiB
| title | parent | nav_order | permalink |
|---|---|---|---|
| Setting up the domain | Installation | 2 | /installation/domainname |
Setting up the domain
Every Matrix server deployment requires a server name which uniquely identifies it. For
example, if you are using the server name example.com, then your users will have usernames
that take the format @user:example.com.
For federation to work, the server name must be resolvable by other homeservers on the internet — that is, the domain must be registered and properly configured with the relevant DNS records.
Matrix servers usually discover each other when federating using the following methods:
- If a well-known delegation exists on
example.com, use the domain and port from the well-known file to connect to the remote homeserver; - If a DNS SRV delegation exists on
example.com, use the IP address and port from the DNS SRV record to connect to the remote homeserver; - If neither well-known or DNS SRV delegation are configured, attempt to connect to the remote
homeserver by connecting to
example.comport TCP/8448 using HTTPS.
The exact details of how server name resolution works can be found in the spec.
TLS certificates
Matrix federation requires that valid TLS certificates are present on the domain. You must obtain certificates from a publicly-trusted certificate authority (CA). Let's Encrypt is a popular choice of CA because the certificates are publicly-trusted, free, and automated via the ACME protocol. (Self-signed certificates are not suitable for federation and will typically not be accepted by other homeservers.)
Automating the renewal of TLS certificates is best practice. There are many tools for this,
but the simplest way to achieve TLS automation is to have your reverse proxy do it for you.
Caddy is recommended as a production-grade reverse proxy with
automatic TLS which is commonly used in front of Dendrite. It obtains and renews TLS certificates
automatically and by default as long as your domain name is pointed at your server first.
Although the finer details of configuring Caddy is not described
here, in general, you must reverse proxy all /_matrix paths to your Dendrite server. For example,
with Caddy:
reverse_proxy /_matrix/* localhost:8008
It is possible for the reverse proxy to listen on the standard HTTPS port TCP/443 so long as your domain delegation is configured to point to port TCP/443.
Delegation
Delegation allows you to specify the server name and port that your Dendrite installation is reachable at, or to host the Dendrite server at a different server name to the domain that is being delegated.
For example, if your Dendrite installation is actually reachable at matrix.example.com port 8448,
you will be able to delegate from example.com to matrix.example.com so that your users will have
@user:example.com user names instead of @user:matrix.example.com usernames.
Delegation can be performed in one of two ways:
- Well-known delegation (preferred): A well-known text file is served over HTTPS on the domain
name that you want to use, pointing to your server on
matrix.example.comport 8448; - DNS SRV delegation (not recommended): See the SRV delegation section below for details.
If you are using a reverse proxy to forward /_matrix to Dendrite, your well-known or delegation
must refer to the hostname and port that the reverse proxy is listening on instead.
Well-known delegation
Using well-known delegation requires that you are running a web server at example.com which
is listening on the standard HTTPS port TCP/443.
Assuming that your Dendrite installation is listening for HTTPS connections at matrix.example.com
on port 8448, the delegation file must be served at https://example.com/.well-known/matrix/server
and contain the following JSON document:
{
"m.server": "matrix.example.com:8448"
}
For example, this can be done with the following Caddy config:
handle /.well-known/matrix/client {
header Content-Type application/json
header Access-Control-Allow-Origin *
respond `{"m.homeserver": {"base_url": "https://matrix.example.com:8448"}}`
}
You can also serve .well-known with Dendrite itself by setting the well_known_server_name config
option to the value you want for m.server. This is primarily useful if Dendrite is exposed on
example.com:443 and you don't want to set up a separate webserver just for serving the .well-known
file.
global:
...
well_known_server_name: "example.com:443"
DNS SRV delegation
This method is not recommended, as the behavior of SRV records in Matrix is rather unintuitive:
SRV records will only change the IP address and port that other servers connect to, they won't
affect the domain name. In technical terms, the Host header and TLS SNI of federation requests
will still be example.com even if the SRV record points at matrix.example.com.
In practice, this means that the server must be configured with valid TLS certificates for
example.com, rather than matrix.example.com as one might intuitively expect. If there's a
reverse proxy in between, the proxy configuration must be written as if it's example.com, as the
proxy will never see the name matrix.example.com in incoming requests.
This behavior also means that if example.com and matrix.example.com point at the same IP
address, there is no reason to have a SRV record pointing at matrix.example.com. It can still
be used to change the port number, but it won't do anything else.
If you understand how SRV records work and still want to use them, the service name is _matrix and
the protocol is _tcp.