For those wishing to more easily integrate [Prometheus](https://prometheus.io/)' alerting service ([Alertmanager](https://prometheus.io/docs/alerting/latest/alertmanager/)) with Matrix, the playbook can now set up [matrix-alertmanager-receiver](https://github.com/metio/matrix-alertmanager-receiver).
See [Setting up Prometheus Alertmanager integration via matrix-alertmanager-receiver](./docs/configuring-playbook-alertmanager-receiver.md) for more details.
**TLDR**: Traefik was migrated from v2 to v3. Minor changes were done to the playbook. Mostly everything else worked out of the box. Most people will not have to do any tweaks to their configuration. In addition, [HTTP/3](https://en.wikipedia.org/wiki/HTTP/3) support is now auto-enabled for the `web-secure` (port 443) and `matrix-federation` (port `8448`) entrypoints. If you have a firewall in front of your server and you wish to benefit from `HTTP3`, you will need to open the `443` and `8448` UDP ports in it.
### Traefik v3
The reverse-proxy that the playbook uses by default (Traefik) has recently been upgraded to v3 (see [this blog post](https://traefik.io/blog/announcing-traefik-proxy-v3-rc/) to learn about its new features). Version 3 includes some small breaking configuration changes requiring a [migration](https://doc.traefik.io/traefik/migration/v2-to-v3/).
We have **updated the playbook to Traefik v3** (make sure to run `just roles` / `make roles` to get it).
There were **only minor playbook changes required** to adapt to Traefik v3, and only to the Ansible role for [matrix-media-repo](./docs/configuring-playbook-matrix-media-repo.md) where we changed a few [`PathPrefix` instances to `PathRegexp`](https://doc.traefik.io/traefik/routing/routers/#path-pathprefix-and-pathregexp), because these instances were using a regular expression instead of a fixed path. For fixed-path values, `PathPrefix` is still the preferred matcher function to use.
**Most people using the playbook should not have to do any changes**.
If you're using the playbook's Traefik instance to reverse-proxy to some other services of your own (not managed by the playbook), you may wish to review their Traefik labels and make sure they're in line with the [Traefik v2 to v3 migration guide](https://doc.traefik.io/traefik/migration/v2-to-v3/).
If you've tweaked any of this playbook's `_path_prefix` variables and made them use a regular expression, you will now need to make additional adjustments. The playbook makes extensive use of `PathPrefix()` matchers in Traefik rules and `PathPrefix` does not support regular expressions anymore. To work around it, you may now need to override a whole `_traefik_rule` variable and switch it from [`PathPrefix` to `PathRegexp`](https://doc.traefik.io/traefik/routing/routers/#path-pathprefix-and-pathregexp).
If you're not using [matrix-media-repo](./docs/configuring-playbook-matrix-media-repo.md) (the only role we had to tweak to adapt it to Traefik v3), you **may potentially downgrade to Traefik v2** (if necessary) by adding `devture_traefik_verison: v2.11.4` to your configuration. People using `matrix-media-repo` cannot downgrade this way, because `matrix-media-repo` has been adjusted to use `PathRegexp` - a [routing matcher](https://doc.traefik.io/traefik/v2.11/routing/routers/#rule) that Traefik v2 does not understand.
### HTTP/3 is enabled by default
In Traefik v3, [HTTP/3](https://en.wikipedia.org/wiki/HTTP/3) support is no longer considered experimental now.
Due to this, **the playbook auto-enables HTTP3** for the `web-secure` (port 443) and `matrix-federation` (port `8448`) entrypoints.
HTTP3 uses the UDP protocol and **the playbook (together with Docker) will make sure that the appropriate ports** (`443` over UDP &`8448` over UDP) **are exposed and whitelisted in your server's firewall**. However, **if you have another firewall in front of your server** (as is the case for many cloud providers), **you will need to manually open these UDP ports**.
If you do not open the UDP ports correctly or there is some other issue, clients (browsers, mostly) will fall-back to [HTTP/2](https://en.wikipedia.org/wiki/HTTP/2) or even [HTTP/1.1](https://en.wikipedia.org/wiki/HTTP).
Still, if HTTP/3 cannot function correctly in your setup, it's best to disable advertising support for it (and misleading clients into trying to use HTTP/3).
To **disable HTTP/3**, you can use the following configuration:
If you are using [your own webserver](./docs/configuring-playbook-own-webserver.md) (in front of Traefik), port binding on UDP port `8448` by default due to HTTP/3 is either unnecessary or [may get in the way](https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/3402). If it does, you can disable it:
```yml
# Disable HTTP/3 for the federation entrypoint.
# If you'd like HTTP/3, consider configuring it for your other reverse-proxy.
#
# Disabling this also sets `matrix_playbook_public_matrix_federation_api_traefik_entrypoint_host_bind_port_udp` to an empty value.
# If you'd like to keep HTTP/3 enabled here (for whatever reason), you may wish to explicitly
# set `matrix_playbook_public_matrix_federation_api_traefik_entrypoint_host_bind_port_udp` to something like '127.0.0.1:8449'.
## synapse-admin is now restricted to your homeserver's URL by default
A new feature introduced in synapse-admin [v0.10.0](https://github.com/Awesome-Technologies/synapse-admin/releases/tag/0.10.0) (released and supported by the playbook since a a few months ago) provides the ability to [restrict its usage to a specific homeserver](https://github.com/Awesome-Technologies/synapse-admin/blob/e21e44362c879ac41f47c580b04210842b6ff3d7/README.md#restricting-available-homeserver) (or multiple homeservers).
The playbook has just started making use of this feature. **From now on, your synapse-admin instance will be restricted to the homeserver you're managing via the playbook**. When configured like this, the *Homeserver URL* field in synapse-admin's web UI changes from a text field to a dropdown having a single value (the URL of your homeserver). This makes usage simpler for most people, as they won't need to manually enter a *Homeserver URL* anymore.
If you'd like **to go back to the old unrestricted behavior**, use the following configuration:
```yml
# Use this configuration to allow synapse-admin to manage any homeserver instance.
## The URL-prefix for Hookshot generic webhooks has changed
Until now, generic Hookshot webhook URLs looked like this: `https://matrix.DOMAIN/hookshot/webhooks/:hookId`.
The `/hookshot/webhooks` common prefix gets stripped by Traefik automatically, so Hookshot only sees the part that comes after (`/:hookId`).
[A few years ago](https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/1681), Hookshot started to prefer to handle webhooks at a `/webhook/:hookId` path (instead of directly at `/:hookId`).
To avoid future problems, we've [reconfigured](https://github.com/spantaleev/matrix-docker-ansible-deploy/commit/4704a60718946fd469aeee7fc3ae8127c633bb6b) our Hookshot configuration to use webhook URLs that include `/webhook` in the URL suffix (e.g. `/hookshot/webhooks/webhook/:hookId`, instead of `/hookshot/webhooks/:hookId`). This means that when we strip the common prefi (`/hookshot/webhooks`), we'll end up sending `/webhook/:hookId` to Hookshot, just like recommended.
When generating new webhooks, you should start seeing the new URLs being used.
**For now**, **both** old URLs (`/hookshot/webhooks/:hookId`) and new URLs (`/hookshot/webhooks/webhook/:hookId`) **continue to work***, so your webhooks will not break just yet.
However, **we recommend that you update all your old webhook URLs** (configured in other systems) to include the new `/webhook` path component, so that future Hookshot changes (whenever they come) will not break your webhooks. You don't need to do anything on the Hookshot side - you merely need to reconfigure the remote systems that use your webhook URLs.
To make things easier and to be consistent with other roles, the [maubot](./docs/configuring-playbook-bot-maubot.md) user (`bot.maubot` by default) is [now](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/3376) automatically created be the playbook.
If you have an existing maubot installation, you will need to specify `matrix_bot_maubot_initial_password` in your `vars.yml` file to make the playbook not complain about it being undefined.
Since the bot is already registered in your installation, there's nothing for the playbook to do anyway. In case you don't remember the password you've registered your maubot user account with, you can specify any value for this variable.
If you've registered another username for the bot (other than the recommended default of `bot.maubot`), consider adjusting the `matrix_bot_maubot_login` variable (e.g. `matrix_bot_maubot_login: my.maubot.username`).
Thanks to [Tobias Diez](https://github.com/tobiasdiez)'s [efforts](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/3241), the playbook now supports bridging to [WeChat](https://www.wechat.com/) via the [matrix-wechat](https://github.com/duo/matrix-wechat) bridge.
See our [Setting up WeChat bridging](docs/configuring-playbook-bridge-wechat.md) documentation page for getting started.
## (Backward Compatibility Break) The playbook now defaults to KeyDB, instead of Redis
**TLDR**: if the playbook used installed Redis as a dependency for you before, it will now replace it with [KeyDB](https://docs.keydb.dev/) (a drop-in alternative) due to [Redis having changed its license](https://redis.com/blog/redis-adopts-dual-source-available-licensing/).
Thanks to [Aine](https://gitlab.com/etke.cc) of [etke.cc](https://etke.cc/), the playbook now uses [KeyDB](https://docs.keydb.dev/) (a drop-in alternative for Redis), instead of [Redis](https://redis.io/).
The playbook used to install Redis (and now installs KeyDB in its place) if services have a need for it ([enabling worker support for Synapse](docs/configuring-playbook-synapse.md#load-balancing-with-workers), [enabling Hookshot encryption](docs/configuring-playbook-bridge-hookshot.md#end-to-bridge-encryption), etc.) or if you explicitly enabled the service (`redis_enabled: true` or `keydb_enabled: true`).
This change is provoked by the fact that [Redis is now "source available"](https://redis.com/blog/redis-adopts-dual-source-available-licensing/). According to the Limitations of [the new license](https://redis.com/legal/rsalv2-agreement/) (as best as we understand them, given that we're not lawyers), using Redis in the playbook (even in a commercial FOSS service like [etke.cc](https://etke.cc/)) does not violate the new Redis license. That said, we'd rather neither risk it, nor endorse shady licenses and products that pretend to be free-software. Another high-quality alternative to Redis seems to be [Dragonfly](https://www.dragonflydb.io/), but the [Dragonfly license](https://github.com/dragonflydb/dragonfly?tab=License-1-ov-file#readme) is no better than Redis's.
Next time your run the playbook (via the `setup-all` tag), **Redis will be automatically uninstalled and replaced with KeyDB**. Some Synapse downtime may occur while the switch happens.
Users on `arm32` should be aware that there's **neither a prebuilt `arm32` container image for KeyDB**, nor the KeyDB role supports self-building yet. Users on this architecture likely don't run Synapse with workers, etc., so they're likely in no need of KeyDB (or Redis). If Redis is necessary in an `arm32` deployment, disabling KeyDB and making the playbook fall back to Redis is possible (see below).
**The playbook still supports Redis** and you can keep using Redis (for now) if you'd like, by adding this additional configuration to your `vars.yml` file:
```yml
# Explicitly disable KeyDB, which will auto-enable Redis
# if the playbook requires it as a dependency for its operation.
Thanks to [Tilo Spannagel](https://github.com/tilosp), the playbook can now enable IPv6 for container networks for various components (roles) via [the `devture_systemd_docker_base_ipv6_enabled` variable](https://github.com/devture/com.devture.ansible.role.systemd_docker_base/blob/c11a526bb8e318b42eb52055056377bb31154f13/defaults/main.yml#L14-L31).
It should be noted that:
- Matrix roles (`roles/custom/matrix-*`) respect this variable, but external roles (those defined in `requirements.yml` and installed via `just roles`) do not respect it yet. Additional work is necessary
- changing the variable subsequently may not change existing container networks. Refer to [these instructions](https://github.com/devture/com.devture.ansible.role.systemd_docker_base/blob/c11a526bb8e318b42eb52055056377bb31154f13/defaults/main.yml#L26-L30)
Thanks to [Julian Foad](https://matrix.to/#/@julian:foad.me.uk), the playbook can now install the [Pantalaimon](https://github.com/matrix-org/pantalaimon) E2EE aware proxy daemon for you. It's already possible to integrate it with [Draupnir](docs/configuring-playbook-bot-draupnir.md) to allow it to work in E2EE rooms - see our Draupnir docs for details.
See our [Setting up Pantalaimon](docs/configuring-playbook-pantalaimon.md) documentation to get started.
Thanks to [FSG-Cat](https://github.com/FSG-Cat), the playbook can now install [Draupnir for all](./docs/configuring-playbook-appservice-draupnir-for-all.md) (aka multi-instance Draupnir running in appservice mode).
This is an alternative to [running Draupnir in bot mode](./docs/configuring-playbook-bot-draupnir.md), which is still supported by the playbook.
The documentation page for [Draupnir for all](./docs/configuring-playbook-appservice-draupnir-for-all.md) contains more information on how to install it.
## Support for bridging to Facebook/Messenger via the new mautrix-meta bridge
The [mautrix-facebook](./docs/configuring-playbook-bridge-mautrix-facebook.md) and [mautrix-instagram](./docs/configuring-playbook-bridge-mautrix-instagram.md) bridges are being [superseded by a new bridge](https://github.com/mautrix/facebook/issues/332) - the [mautrix-meta](https://github.com/mautrix/meta) bridge.
The playbook now supports the new mautrix-meta bridge - a single bridge, which can run in different modes and bridge to Messenger (via [Facebook](https://facebook.com/), Facebook over [Tor](https://www.torproject.org/) or via [Messenger](https://messenger.com/)) and [Instagram](https://instagram.com/). The playbook makes this bridge available via 2 separate Ansible roles, allowing you to easily run 2 instances of mautrix-meta, for bridging to both services at the same time.
If you're using mautrix-facebook or mautrix-instagram right now, **you can still continue using the old bridges, but may wish to change to the new bridge implementations**. See:
- [Setting up Instagram bridging via Mautrix Meta](docs/configuring-playbook-bridge-mautrix-meta-instagram.md)
- [Setting up Messenger bridging via Mautrix Meta](docs/configuring-playbook-bridge-mautrix-meta-messenger.md)
The documentation pages contain more information on how to migrate.
## Much larger Synapse caches and cache auto-tuning enabled by default
Thanks to [FSG-Cat](https://github.com/FSG-Cat), the playbook now uses much larger caches and enables Synapse's [cache auto-tuning functionality](https://matrix-org.github.io/synapse/latest/usage/configuration/config_documentation.html#caches-and-associated-values).
This work and the default values used by the playbook are inspired by [Tom Foster](https://github.com/tcpipuk)'s [Synapse homeserver guide](https://tcpipuk.github.io/synapse/deployment/synapse.html).
The playbook has always used a very conservative cache factor (`matrix_synapse_caches_global_factor`) value of `0.5`, which may be OK for small and underactive deployments, but is not ideal for larger servers. Paradoxically, a small global cache factor value [does not necessarily decrease RAM usage as a whole](https://github.com/matrix-org/synapse/issues/3939).
The playbook now uses **a 20x larger cache factor** (currently `10`), adjusts a few other cache-related variables, and **enables cache auto-tuning** via the following variables:
-`matrix_synapse_cache_autotuning_min_cache_ttl` - defaults to `30s`
These values should be good defaults for most servers, but may change over time as we experiment further.
Refer to our new [Tuning caches and cache autotuning](docs/maintenance-synapse.md#tuning-caches-and-cache-autotuning) documentation section for more details.
## (Backward-compatibility break) Minor changes necessary for some people serving a static website at the base domain
This only affects people who are [Serving a static website at the base domain](./docs/configuring-playbook-base-domain-serving.md#serving-a-static-website-at-the-base-domain), but not managing its `index.html` through the playbook.
That is, for people who have `matrix_static_files_file_index_html_enabled: false` in their `vars.yml` configuration, the playbook has a new default behavior. Since the playbook is not managing the `index.html` file, it will default to a more sensible way of handling the base domain - redirecting `https://DOMAIN/` to `https://matrix.DOMAIN/`, instead of serving a 404 page.
If you are managing your static website by yourself (by dropping files into `/matrix/static-files/public` somehow), then you probably don't wish for such redirection to happen. You can disable it by adding `matrix_static_files_container_labels_base_domain_root_path_redirection_enabled: false` to your `vars.yml` configuration file.
Thanks to [Charles Wright](https://github.com/cvwright) from [FUTO](https://www.futo.org/), the creators of the [Circles app](https://circu.li/), the playbook has [received support](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/3100) for load-balancing the Synapse workload via [specialized workers](./docs/configuring-playbook-synapse.md#specialized-workers) which are supposed to work better than our old [generic workers](./docs/configuring-playbook-synapse.md#generic-workers) implementation.
For now, playbook defaults remain unchanged and the `one-of-each` [workers preset](./docs/configuring-playbook-synapse.md#worker-presets) continues being the default. However, the default may change in the future. If you'd like to remain on this preset even if/when the defaults change, consider explicitly adding `matrix_synapse_workers_preset: one-of-each` to your `vars.yml` configuration.
Our specialized workers setup is based on recommendations found in [Tom Foster](https://github.com/tcpipuk)'s [Synapse homeserver guide](https://tcpipuk.github.io/synapse/index.html). What's special about our new setup is that we try to parse information out of the request (who the user is; which room is being operated on) and try to forward similar requests to the same worker. As an example, this means that once a worker caches some room information, subsequent requests for the same room will be routed to the same worker (which supposedly still has the room's state cached).
To get started, refer to our [Specialized workers](./docs/configuring-playbook-synapse.md#specialized-workers) documentation section.
## Switching to Element's AGPLv3-licensed Synapse release
A few months ago, the [Element](https://element.io/) company has [announced](https://element.io/blog/element-to-adopt-agplv3/) that their work on the Synapse homeserver would no longer be available under the permissive [Apache-2.0 license](https://www.apache.org/licenses/LICENSE-2.0), but only under:
- the [AGPLv3](https://www.gnu.org/licenses/agpl-3.0.en.html) free-software license - the same license that this Ansible playbook has always used
- a proprietary license, for those wishing for Element to [sell them an exception](https://gnu.org/philosophy/selling-exceptions.html) to the AGPLv3 license
You can also learn more in [this post](https://matrix.org/blog/2023/11/06/future-of-synapse-dendrite/) by the Matrix Foundation.
The change has [already happened](https://element.io/blog/synapse-now-lives-at-github-com-element-hq-synapse/) and the first Synapse release under the new license is here: [v1.99.0](https://github.com/element-hq/synapse/releases/tag/v1.99.0).
There is no up-to-date alternative Synapse fork right now and this free-software (AGPLv3-licensed) playbook is definitely not against free-software licenses, so we are now switching to the Element-maintained Synapse release.
**What does this mean to you?**
For most home users, it doesn't mean anything. Your installation will continue working as it should and you don't need to do anything.
For people building commercial products on top of Synapse, they may have to either buy a license exception from Element (from what we hear, the fee depends on the number of monthly-active users on your instance) or they may need to release all related code as free-software (which is what we've been doing at [etke.cc](https://etke.cc/) ([here](https://gitlab.com/etke.cc)) all along).
We're no lawyers and this changelog entry does not aim to give you the best legal advice, so please research on your own!
If you'd like to continue using the old Apache-2.0-licensed Synapse (for a while longer anyway), the playbook makes it possible by intruducing a new Ansible variable. You can do it like this:
```yaml
# Switch the organization that Synapse container images (or source code for self-building) are pulled from.
# Note: the new default value is `element-hq/synapse`.
# Pin the Synapse version to the last one (v1.98.0) released by the Matrix Foundation
# under the old permissive Apache-2.0 license.
matrix_synapse_version: v1.98.0
```
Notes:
- if you had already upgraded Synapse to `v1.99.0` by running this playbook, you will still be able to downgrade to `v1.98.0`, because both releases use the same database schema version (`SCHEMA_COMPAT_VERSION = 83` - see [here for v1.98.0](https://github.com/element-hq/synapse/blob/v1.98.0/synapse/storage/schema/__init__.py#L131-L134) and [here for v1.99.0](https://github.com/element-hq/synapse/blob/v1.99.0/synapse/storage/schema/__init__.py#L137-L140)). More details on Synapse's database schema are available [here](https://element-hq.github.io/synapse/develop/development/database_schema.html). It appears that there are no new database migrations introduced in `v1.99.0`, so going back to the older release is possible. This is not guaranteed to hold true for future Synapse releases, so if you're seeing this early-enough, consider pinning the version and organization before re-running the playbook and getting upgraded to the latest version
- running an outdated homeserver exposes you to security issues and incompatibilities. Only consider doing this as a short-term solution.
As of [#204](https://github.com/the-draupnir-project/Draupnir/pull/204) Draupnir changed its licence to AFL-3.0 from the CSL licence. This change affects playbook users who could not run Draupnir under the old license restrictions. The new license is considerably less restrictive and is OSI approved. Draupnir version v1.86.0 and later are covered by this license change.
**TLDR**: All traces of the `matrix-nginx-proxy` reverse-proxy component are now gone. This brought about many other internal changes (and security improvements), so setups may need minor adjustments or suffer some (temporary) breakage. People who have been on the Traefik-native setup may upgrade without much issues. Those running their own Traefik instance may need minor changes. People who have been postponing the migration away from `matrix-nginx-proxy` (for more than a year already!) will now finally need to do something about it.
### Backstory on `matrix-nginx-proxy`
We gather here today to celebrate the loss of a once-beloved component in our stack - `matrix-nginx-proxy`. It's been our [nginx](https://nginx.org/)-based reverse-proxy of choice since the [first commit](https://github.com/spantaleev/matrix-docker-ansible-deploy/tree/87f5883f2455fb115457b65f267f17de305c053c) of this playbook, 7 years ago.
For 6 years, `matrix-nginx-proxy` has been the front-most reverse-proxy in our setup (doing SSL termination, etc.). After [transitioning to Traefik last year](#traefik-is-the-default-reverse-proxy-now), `matrix-nginx-proxy` took a step back. Nevertheless, since it was so ingrained into the playbook, it still remained in use - even if only internally. Despite our warnings of its imminent death, many of you have indubitably continued to use it instead of Traefik. Its suffering continued for too long, because it served many different purposes and massive effort was required to transition them to others.
To us, `matrix-nginx-proxy` was:
- an [nginx](https://nginx.org/)-based reverse-proxy
- an Ansible role organizing the work of [certbot](https://certbot.eff.org/) - retrieving free [Let's Encrypt](https://letsencrypt.org/) SSL certificates for `matrix-nginx-proxy` and for the [Coturn TURN server](./docs/configuring-playbook-turn.md)
- a central component for reverse-proxying to the [long list of services](./docs/configuring-playbook.md) supported by the playbook. As such, it became a dependency that all these services had to inject themselves into during runtime
- an intermediary through which addons (bridges, bots) communicated with the homeserver. Going through an intermediary (instead of directly talking to the homeserver) is useful when certain components (like [matrix-media-repo](./docs/configuring-playbook-matrix-media-repo.md) or [matrix-corporal](./docs/configuring-playbook-matrix-corporal.md)) are enabled, because it lets these services "steal routes" from the homeserver
- a webserver for serving the `/.well-known/matrix` static files (generated by the `matrix-base` role until now)
- a webserver [serving your base domain](./docs/configuring-playbook-base-domain-serving.md) (and also generating the `index.html` page for it)
- a central component providing global [HTTP Basic Auth](https://developer.mozilla.org/en-US/docs/Web/HTTP/Authentication) password-protection for all `/metrics` endpoints when metrics were exposed publicly for consumption from a remote Prometheus server
Talk about a jack of all trades! The [UNIX philosophy](https://en.wikipedia.org/wiki/Unix_philosophy) (and Docker container philosophy) of "do one thing and do it well" had been severely violated for too long.
On a related note, we also had a large chain of reverse-proxies in the mix.
In the worst case, it was something like this: (Traefik -> `matrix-nginx-proxy:8080` -> `matrix-nginx-proxy:12080` -> `matrix-synapse-reverse-proxy-companion:8008` -> `matrix-synapse:8008`).
Due to complexity and the playbook's flexibility (trying to accommodate a mix of tens of components), many layers of indirection were necessary. We do like reverse-proxies, but.. not quite enough to enjoy going through a chain of ~4 of them before reaching the target service.
After **a ton of work** in the last weeks (200+ commits, which changed 467 files - 8684 insertions and 8913 deletions), **we're finally saying goodbye** to `matrix-nginx-proxy`.
### Going Traefik-native and cutting out all middlemen
In our new setup, you'll see the bare minimum number of reverse-proxies.
In most cases, there's only Traefik and all services being registered directly with it. When [Synapse workers](./docs/configuring-playbook-synapse.md#load-balancing-with-workers) are enabled, `matrix-synapse-reverse-proxy-companion` remains as an extra reverse-proxy that requests go through (for load-balancing to the correct Synapse worker), but in all other cases services are exposed directly.
This reduces "network" hops (improving performance) and also decreases the number of components (containers).
Each Ansible role in our setup is now independent and doesn't need to interact with other roles during runtime.
### Traefik now has an extra job
Previously, **Traefik had a single purpose** - being the main reverse-proxy. It was either front-most (terminating SSL, etc.) or you were [fronting Traefik with your own other reverse-proxy](./docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy). In any case - it had this central (yet decentralized) job.
Now, **Traefik has one more role** - it serves as an intermediary which allows addon services (bridges, bots, etc.) to communicate with the homeserver. As mentioned above, such an intermediary service is not strictly necessary in all kinds of setups, but more complex setups (including [matrix-media-repo](./docs/configuring-playbook-matrix-media-repo.md) or [matrix-corporal](./docs/configuring-playbook-matrix-corporal.md)) benefit from it.
To perform this new role, Traefik now has a new internal [entrypoint](https://doc.traefik.io/traefik/routing/entrypoints/) called `matrix-internal-matrix-client-api`. All homeservers (Conduit, Dendrite, Synapse and even `matrix-synapse-reverse-proxy-companion`) and homeserver-related core services ([matrix-media-repo](./docs/configuring-playbook-matrix-media-repo.md), [matrix-corporal](./docs/configuring-playbook-matrix-corporal.md) and potentially others) register their routes (using [container labels](https://docs.docker.com/config/labels-custom-metadata/)) not only on the public entrypoints (`web-secure`, `matrix-federation`), but also on this new internal entrypoint.
Doing so, services can contact Traefik on this entrypoint's dedicated port (the URL defaults to `http://matrix-traefik:8008`) and reach the homeserver Client-Server API as they expect. Internally, Traefik takes care of the routing to the correct service.
We've also considered keeping it simple and having services talk to the homeserver over the public internet (e.g. `https://matrix.DOMAIN`) thus reusing all existing Traefik routing labels. In this scenario, performance was incredibly poor (e.g. 70 rps, instead of 1400 rps) due to TLS and networking overhead. The need for fast internal communication (via the new internal non-TLS-enabled Traefik entrypoint) is definitely there. In our benchmarks, Traefik even proved more efficient than nginx at doing this: ~1200 rps for Traefik compared to ~900 rps for nginx (out of ~1400 rps when talking to the Synapse homeserver directly).
Traefik serving this second purpose has a few downsides:
- Traefik becomes a runtime dependency for all homeserver-dependant container services
- all homeserver-dependant services now need to be connected to the `traefik` container network, even if they don't need public internet exposure
Despite these downsides (which the playbook manages automatically), we believe it's still a good compromise given the amount of complexity it eliminates and the performance benefits it yields. One alternative we've [considered](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/3045#issuecomment-1867327001) was adding a new intermediary service (e.g. `matrix-homeserver-proxy` powered by nginx), but this both had much higher complexity (one more component in the mix; duplication of effort to produce nginx-compatible route definitions for it) and slightly worse performance (see above).
People running the default Traefik setup do not need to do anything to make Traefik take on this extra job. Your Traefik configuration will be updated automatically.
**People runnning their own Traefik reverse-proxy need to do [minor adjustments](#people-managing-their-own-traefik-instance-need-to-do-minor-changes)**, as described in the section below.
You may disable Traefik acting as an intermediary by explicitly setting `matrix_playbook_public_matrix_federation_api_traefik_entrypoint_enabled` to `false`. Services would then be configured to talk to the homeserver directly, giving you a slight performance boost and a "simpler" Traefik setup. However, such a configuration is less tested and will cause troubles, especially if you enable more services (like `matrix-media-repo`, etc.) in the future. As such, it's not recommended.
### People managing their own Traefik instance need to do minor changes
This section is for people [managing their own Traefik instance on the Matrix server](./docs/configuring-playbook-own-webserver.md#traefik-managed-by-you). Those [using Traefik managed by the playbook](./docs/configuring-playbook-own-webserver.md#traefik-managed-by-the-playbook) don't need to do any changes.
Because [Traefik has an extra job now](#traefik-now-has-an-extra-job), you need to adapt your configuration to add the additional `matrix-internal-matrix-client-api` entrypoint and potentially configure the `matrix_playbook_reverse_proxy_container_network` variable. See the [Traefik managed by you](./docs/configuring-playbook-own-webserver.md#traefik-managed-by-you) documentation section for more details.
### People fronting Traefik with another reverse proxy need to do minor changes
We've already previously mentioned that you need to do some minor [configuration changes related to `devture_traefik_additional_entrypoints_auto`](#backward-compatibility-configuration-changes-required-for-people-fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy).
If you don't do these changes (switching from `devture_traefik_additional_entrypoints_auto` to multiple other variables), your Traefik setup will not automatically receive the new `matrix-internal-matrix-client-api` Traefik entrypoint and Traefik would not be able to perform [its new duty of connecting addons with the homeserver](#traefik-now-has-an-extra-job).
### Supported reverse proxy types are now fewer
This section is for people using a more custom reverse-proxy setup - those having `matrix_playbook_reverse_proxy_type` set to a value different than the default (`playbook-managed-traefik`).
Previously, we allowed you to set `matrix_playbook_reverse_proxy_type` to 7 different values to accommodate various reverse-proxy setups.
The complexity of this is too high, so we only support 3 values right now:
- (the default) `playbook-managed-traefik`, when you're [using Traefik managed by the playbook](./docs/configuring-playbook-own-webserver.md#traefik-managed-by-the-playbook)
-`other-traefik-container`, when you're [managing your own Traefik instance on the Matrix server](./docs/configuring-playbook-own-webserver.md#traefik-managed-by-you)
-`none`, when you wish for [no reverse-proxy integration to be done at all](./docs/configuring-playbook-own-webserver.md#using-no-reverse-proxy-on-the-matrix-side-at-all)
The `none` value is not recommended and may not work adequately, due to lack of testing and [Traefik's new responsibilities](#traefik-now-has-an-extra-job) in our setup.
**Previous values that are now gone** (and the playbook would report them as such) are: `playbook-managed-nginx`, `other-nginx-non-container`, `other-on-same-host` and `other-on-another-host`.
If you were using these values as a way to stay away from Traefik, you now have 2 options:
- (recommended) [Fronting Traefik with another reverse-proxy](./docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy)
- (not recommended) [Using no reverse-proxy on the Matrix side at all](./docs/configuring-playbook-own-webserver.md#using-no-reverse-proxy-on-the-matrix-side-at-all) and reverse-proxying to each and every service manually
### Container networking changes
Now that `matrix-nginx-proxy` is not in the mix, it became easier to clear out some other long-overdue technical debt.
Since the very beginning of this playbook, all playbook services were connected to a single (shared) `matrix` container network. Later on, some additional container networks appeared, but most services (database, etc.) still remained in the `matrix` container network. This meant that any random container in this network could try to talk (or attack) the Postgres database operating in the same `matrix` network.
Moving components (especially the database) into other container networks was difficult - it required changes to many other components to ensure correct connectivity.
All the hard work has been done now. We've added much more isolation between services by splitting them up into separate networks (`matrix-homeserver`, `matrix-addons`, `matrix-monitoring`, `matrix-exim-relay`, etc). Components are only joined to the networks they need and should (for the most part) not be able to access unrelated things.
Carrying out these container networking changes necessitated modifying many components, so **we're hoping not too many bugs were introduced in the process**.
We've refrained from creating too many container networks (e.g. one for each component), to avoid exhausting Docker's default network pool and contaminating the container networks list too much.
### Metrics exposure changes
This section is for people who are exposing monitoring metrics publicly, to be consumed by an external Prometheus server.
Previously, `matrix-nginx-proxy` was potentially password-protecting all `/metrics/*` endpoints with the same username and password (specified as plain-text in your `vars.yml` configuration file).
From now on, there are new variables for doing roughly the same - `matrix_metrics_exposure_enabled`, `matrix_metrics_exposure_http_basic_auth_enabled` and `matrix_metrics_exposure_http_basic_auth_users`. See the [Prometheus & Grafana](./docs/configuring-playbook-prometheus-grafana.md) docs page for details.
`matrix-nginx-proxy` is not acting as a "global guardian" anymore. Now, each role provides its own metrics exposure and protection by registering with Traefik. Nevertheless, all roles are wired (via playbook configuration in `group_vars/matrix_servers`) to obey these new `matrix_metrics_exposure_*` variables. We've eliminated the centralization, but have kept the ease of use. Now, you can also do per-service password-protection (with different credentials), should you need to do that for some reason.
The playbook will tell you about all variables that you need to migrate during runtime, so rest assured - you shouldn't be able to miss anything!
### Matrix static files
As mentioned above, static files like `/.well-known/matrix/*` or your base domain's `index.html` file (when [serving the base domain via the Matrix server](./docs/configuring-playbook-base-domain-serving.md) was enabled) were generated by the `matrix-base` or `matrix-nginx-proxy` roles and put into a `/matrix/static-files` directory on the server. Then `matrix-nginx-proxy` was serving all these static files.
All of this has been extracted into a new `matrix-static-files` Ansible role that's part of the playbook. The static files generated by this new role still live at roughly the same place (`/matrix/static-files/public` directory, instead of `/matrix/static-files`).
The playbook will migrate and update the `/.well-known/matrix/*` files automatically but not your own files in `nginx-proxy/data/matrix-domain/` you will need to back these up yourself otherwise they will be lost. It will also warn you about usage of old variable names, so you can adapt to the new names.
Some of you have been voicing their concerns (for a long time) about Traefik being too slow and nginx being better.
Some online benchmarks support this by demonstrating slightly higher SSL-termination performance in favor of nginx. The upcoming Traefik v3 release is [said to](https://medium.com/beyn-technology/is-nginx-dead-is-traefik-v3-20-faster-than-traefik-v2-f28ffb7eed3e) improve Traefik's SSL performance by some 20%, but that still ends up being somewhat slower than nginx.
We believe that using Traefik provides way too many benefits to worry about this minor performance impairment.
The heaviest part of running a Matrix homeserver is all the slow and potentially inefficient things the homeserver (e.g. Synapse) is doing. These things affect performance much more than whatever reverse-proxy is in front. Your server will die the same way by joining the famously large **Matrix HQ** room, no matter which reverse-proxy you put in front.
Even our previously mentioned benchmarks (yielding ~1300 rps) are synthetic - hitting a useless `/_matrix/client/versions` endpoint. Real-use does much more than this.
If this is still not convincing enough for you and you want the best possible performance, consider [Fronting Traefik with another reverse-proxy](./docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy) (thus having the slowest part - SSL termination - happen elsewhere) or [Using no reverse-proxy on the Matrix side at all](./docs/configuring-playbook-own-webserver.md#using-no-reverse-proxy-on-the-matrix-side-at-all). The playbook will not get in your way of doing that, but these options may make your life much harder. Performance comes at a cost, after all.
### Migration procedure
The updated playbook will automatically perform some migration tasks for you:
1. It will stop and remove the `matrix-nginx-proxy` systemd service and container for you. This behavior cannot be disabled. It's essential that this service gets stopped, because it remaining running (and having container labels) may confuse Traefik as to where to route HTTP requests.
2. It will delete the `/matrix/nginx-proxy` directory and all files within it. You can disable this behavior by adding `matrix_playbook_migration_matrix_nginx_proxy_uninstallation_enabled: false` to your `vars.yml` configuration file. Doing so will leave its data around.
3. It will delete the `/matrix/ssl` directory and all files within it. You can disable this behavior by adding `matrix_playbook_migration_matrix_ssl_uninstallation_enabled: false` to your `vars.yml` configuration file. If you have some important certificates there for some reason, take them out or temporarily disable removal of these files until you do.
4. It will tell you about all variables (`matrix_nginx_proxy_*` and many others - even from other roles) that have changed during this large nginx-elimination upgrade. You can disable this behavior by adding `matrix_playbook_migration_matrix_nginx_proxy_elimination_variable_transition_checks_enabled: false` to your `vars.yml` configuration file.
5. It will tell you about any leftover `matrix_nginx_proxy_*` variables in your `vars.yml` file. You can disable this behavior by adding `matrix_playbook_migration_matrix_nginx_proxy_leftover_variable_validation_checks_enabled: false` to your `vars.yml` configuration file.
6. It will tell you about any leftover `matrix_ssl_*` variables in your `vars.yml` file. You can disable this behavior by adding `matrix_playbook_migration_matrix_ssl_leftover_variable_checks_enabled: false` to your `vars.yml` configuration file.
We don't recommend changing these variables and suppressing warnings, unless you know what you're doing.
**Most people should just upgrade as per-normal**, bearing in mind that a lot has changed and some issues may arise.
The playbook would guide you through renamed variables automatically.
### Conclusion
Thousands of lines of code were changed across hundreds of files.
All addons (bridges, bots) were rewired in terms of container networking and in terms of how they reach the homeserver.
I don't actively use all the ~100 components offered by the playbook (no one does), nor do I operate servers exercising all edge-cases. As such, issues may arise. Please have patience and report (or try to fix) these issues!
## (Backward Compatibility) Configuration changes required for people fronting the integrated reverse-proxy webserver with another reverse-proxy
If you're on the default setup (using the Traefik reverse-proxy as installed by the playbook), you don't need to do anything.
People who are [Fronting the integrated Traefik reverse-proxy webserver with another reverse-proxy](./docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy), as per our previous instructions are redefining `devture_traefik_additional_entrypoints_auto` in their `vars.yml` configuration.
Such a full variable redefinion is intrustive, because it prevents the playbook from injecting additional entrypoints into the Traefik webserver. In the future, the playbook may have a need to do so.
For this reason, we no longer recommend completely redefining `devture_traefik_additional_entrypoints_auto`.
The playbook now defines [various `matrix_playbook_public_matrix_federation_api_traefik_entrypoint_*` variables in the `defaults/main.yml` file](https://github.com/spantaleev/matrix-docker-ansible-deploy/blob/master/roles/custom/matrix-base/defaults/main.yml) of the `matrix-base` role which can be used as a safer alternative to `devture_traefik_additional_entrypoints_auto`.
Adapt your configuration as seen below:
```diff
-devture_traefik_additional_entrypoints_auto:
- - name: matrix-federation
- port: 8449
- host_bind_port: '127.0.0.1:8449'
- config: {}
- # If your reverse-proxy runs on another machine, remove the config above and use this config instead:
Also, feel free to read the [Fronting the integrated Traefik reverse-proxy webserver with another reverse-proxy](./docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy) documentation section again for additional details.
## matrix-reminder-bot update with more secure (backward-incompatible) default settings
**TLDR**: your updated (to [v0.3.0](https://github.com/anoadragon453/matrix-reminder-bot/releases/tag/v0.3.0)) [matrix-reminder-bot](./docs/configuring-playbook-bot-matrix-reminder-bot.md) is now more secure. By default, like other bridges/bots managed by the playbook, it will only provide its services to users of your own server (not to anyone, even across the Matrix Federation). If that's fine, there's nothing you need to do.
Maintenance of [matrix-reminder-bot](./docs/configuring-playbook-bot-matrix-reminder-bot.md) has been picked up by [Kim Brose](https://github.com/HarHarLinks) and [@svierne](https://github.com/svierne).
Thanks to them, a new [v0.3.0](https://github.com/anoadragon453/matrix-reminder-bot/releases/tag/v0.3.0) release is out. The new version is now available for the ARM64 architecture, so playbook users on this architecture will no longer need to wait for [self-building](./docs/self-building.md) to happen.
The new version also comes with new `allowlist` and `blocklist` settings, which make it possible to restrict who can use the bot. Previously anyone, even across the Matrix Federation could talk to it and schedule reminders.
The playbook defaults all bridges and bots (where possible) to only be exposed to users of the current homeserver, not users across federation.
Thanks to the new version of this bot making such a restriction possible, we're now making use of it. The playbook (via its `group_vars/matrix_servers` file) automatically enables the `allowlist` (`matrix_bot_matrix_reminder_bot_allowlist_enabled: true`) and configures it in such a way (`matrix_bot_matrix_reminder_bot_allowlist_regexes_auto`) so as to restrict the bot to your homeserver's users.
If you need **to undo or tweak these security improvements**, you can change your `vars.yml` file to:
- disable the allowlist (`matrix_bot_matrix_reminder_bot_allowlist_enabled: false`), making the bot allow usage by anyone, anywhere
- inject additional allowed servers or users by adding **additional** (on top of the default allowlist in `matrix_bot_matrix_reminder_bot_allowlist_regexes_auto`) custom regexes in the `matrix_bot_matrix_reminder_bot_allowlist_regexes_custom` list variable (see the [syntax reference](https://github.com/anoadragon453/matrix-reminder-bot/blob/1e910c0aa3469d280d93ee7e6c6d577227a3460c/sample.config.yaml#L43-L49))
- override the default allowlist (in the `group_vars/matrix_servers` file) by redefining `matrix_bot_matrix_reminder_bot_allowlist_regexes_auto`
## matrix-mailer has been replaced by the exim-relay external role
We're continuing our effort to make [the playbook use external roles for some things](#the-playbook-now-uses-external-roles-for-some-things), so as to avoid doing everything ourselves and to facilitate code re-use.
The `matrix-mailer` role has been moved to its own repository ([ansible-role-exim-relay](https://github.com/mother-of-all-self-hosting/ansible-role-exim-relay)) that this playbook now includes.
To migrate:
- pull the playbook changes, as usual
- update your roles (run `just roles` or `make roles`)
- update your `vars.yml`, renaming `matrix_mailer`-prefixed variables to `exim_relay`-prefixed ones (e.g. `matrix_mailer_sender_address` -> `exim_relay_sender_address`). If you find none, it means you're using the default configuration and your migraiton job is even simpler.
- re-run the playbook (`install-all` or `setup-all`)
The playbook will take care of stopping the old `matrix-mailer` systemd service, relocating its directory and restarting it under the new name (`matrix-exim-relay.service`).
## mautrix-signal now powered by the new Go-based bridge
The old Python-based [mautrix-signal](https://github.com/mautrix/signal) bridge is no longer maintained upstream. It's also known to have issues linking new devices.
It seems like the path forward is to switch to the new mautrix-signal bridge written in Golang, which we did thanks to [PR #3031](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/3041) by [Pierre 'McFly' Marty](https://github.com/pm-McFly).
The playbook should **automatically migrate your mautrix-signal installation to the new bridge code**.
You will **need to relink all your devices** to continue your bridged conversations.
**TDLR**: if your Matrix server is federating (which it mostly likely is, unless you've [disabled federation](docs/configuring-playbook-federation.md#disabling-federation)), your public rooms will not only be joinable across federation (as they've always been), but from now on will be discoverable (made available as a list across federation). We're changing this by flipping the value for Synapse's `allow_public_rooms_over_federation` setting to `true`, going against the upstream default. Servers that disable federation are not affected. Servers that have public rooms which are not published to the room directory are also not affected.
We generally try to stick to the default configuration for Synapse (and all other components), unless these defaults seem wrong or harmful. One such previous case from a few months ago was us [Enabling `forget_rooms_on_leave` by default for Synapse](#enabling-forget_rooms_on_leave-by-default-for-synapse) - the default value was making Synapse more wasteful of resources by default.
Today, we're going against upstream defaults again and flipping the `allow_public_rooms_over_federation` configuration option to `true`.
This way, public rooms on your server will be made discoverable by others via federation, using the [`GET /_matrix/federation/v1/publicRooms` of the Server-Server API](https://spec.matrix.org/v1.8/server-server-api/#get_matrixfederationv1publicrooms).
The upstream Synapse default is `false` (disabled), so that public rooms are not exposed for other servers to discover (learn about their existence). Nevertheless, even if these rooms are not exposed (listed) for discovery, they are **still joinable** by anyone who knows their address or is invited to the room by an existing member.
**We go against the upstream default** in an effort to make Matrix federation more useful - a public room should be globally public - not only joinable, but also discoverable across federation.
The **historical reasoning** behind this change is as follows:
-`allow_public_rooms_over_federation` seems to have been enabled by default for Synapse until v1.7.0 (~2019), just like we believe it should be for a globally-federating network - rooms should be joinable and discoverable across federation.
- In Synapse v1.7.0 (~2019), `allow_public_rooms_over_federation` [got disabled](https://github.com/element-hq/synapse/blob/e9069c9f919685606506f04527332e83fbfa44d9/docs/upgrade.md?plain=1#L1877-L1891) by default in a [security-by-obscurity](https://en.wikipedia.org/wiki/Security_through_obscurity) workaround for misconfigured servers. See the [Avoiding unwelcome visitors on private Matrix servers](https://matrix.org/blog/2019/11/09/avoiding-unwelcome-visitors-on-private-matrix-servers/) `matrix.org` blog article. We believe that people wishing for a truly private server, should [disable federation](docs/configuring-playbook-federation.md#disabling-federation), instead of having a fully-federating server and trying to hide its public rooms. We also provide other workarounds below. We (and the Synapse team, obviously) believe that Matrix should federate by default, so federating the public room list seems to make sense.
- [etke.cc](https://etke.cc/) has been developing the free-software [Matrix Rooms Search](https://gitlab.com/etke.cc/mrs) project for a while now. One public (demo) instance of it is hosted at [matrixrooms.info](https://matrixrooms.info/). This search engine tries to go through the Matrix federation and discover & index public rooms to allow people to find them. We believe it's vital for Matrix (and any chat or social network for that matter) to be more discoverable, so that people can find communities and others to talk to. Today (on 23rd of October 2023), `matrixrooms.info` is indexing `23066` Matrix servers. Of these, only `1567` servers (7%) are making their public rooms discoverable. Who knows what wonderful communities and rooms are available on these 93% other Matrix servers that are supposedly federating, but are still gate-keeping their public room list. Indubitably, many of these servers are hosted via matrix-docker-ansible-deploy, so we feel partially responsible for making Matrix federation less useful.
Here are **actions you may wish to take** as a result of this change:
- (recommended) embrace the new default. If your Matrix server is federating, your public rooms have always been joinable across federation anyway. Exposing the list of public rooms does no harm and more-so does good by contributing to the usefulness of the Matrix network by facilitating room discovery.
- (switch to a better way of doings things on your semi-private server) The problem that the Synapse team appears to have solved by flipping the `allow_public_rooms_over_federation` default in Synapse v1.7.0 seems to for "mostly private" servers, which federate and have a bunch of rooms made public (and published in their room directory) in an effort to allow people on the same homeserver to easily find and join them (self-onboarding). With the introduction of Matrix Spaces, you can reorganize your flow around spaces - you can auto-join your users to a Matrix Space (via Synapse's `auto_join_rooms` setting - controlled by our `matrix_synapse_auto_join_rooms` variable), then add a bunch of rooms to the space and make them joinable by people belonging to the space. That is to say, do not make rooms public and do not publish them to the room directory unless they are really public. Instead, use other mechanisms for semi-public rooms or private rooms. One alternative is to stick to what you're doing (public rooms published to your rooms directory) but having a `m.federate: true` flag set during creation (clients like Element have a nice UI checkbox for this) to explicitly disable federation for them.
- (keeping the old behavior) if you wish to keep doing what you're doing (keeping your Matrix server federating, but hiding its public rooms list), add `matrix_synapse_allow_public_rooms_over_federation: false` to your `vars.yml` configuration. This restores the old behavior. You may also consider [disabling federation](docs/configuring-playbook-federation.md#disabling-federation) completely instead of relying on security-by-obscurity measures.
## Postgres parameters are automatically tuned now
The playbook has provided some hints about [Tuning PostgreSQL](docs/maintenance-postgres.md#tuning-postgresql) for quite a while now.
From now on, the [Postgres Ansible role](https://github.com/devture/com.devture.ansible.role.postgres) automatically tunes your Postgres configuration with the same [calculation logic](https://github.com/le0pard/pgtune/blob/master/src/features/configuration/configurationSlice.js) that powers https://pgtune.leopard.in.ua/.
Our [Tuning PostgreSQL](docs/maintenance-postgres.md#tuning-postgresql) documentation page has details about how you can turn auto-tuning off or adjust the automatically-determined Postgres configuration parameters manually.
People who [enable load-balancing with Synapse workers](docs/configuring-playbook-synapse.md#load-balancing-with-workers) no longer need to increase the maximum number of Postgres connections manually (previously done via `devture_postgres_process_extra_arguments`). There's a new variable (`devture_postgres_max_connections`) for controlling this number and the playbook automatically raises its value from `200` to `500` for setups which enable workers.
Thanks to [Aine](https://gitlab.com/etke.cc) of [etke.cc](https://etke.cc/), the playbook can now set up the [SchildiChat](https://github.com/SchildiChat/schildichat-desktop) client.
See our [Configuring SchildiChat](docs/configuring-playbook-client-schildichat.md) documentation to get started.
See our [Setting up Mautrix wsproxy for bridging Android SMS or Apple iMessage](docs/configuring-playbook-bridge-mautrix-wsproxy.md) documentation page for getting started.
[matrix-registration-bot](docs/configuring-playbook-bot-matrix-registration-bot.md) got some updates and now supports password-only-based login. Therefore the bot now doesn't need any manual configuration except setting a password in your `vars.yml`. The bot will be registered as admin and access tokens will be obtained automatically by the bot.
**For existing users** You need to set `matrix_bot_matrix_registration_bot_bot_password` if you previously only used `matrix_bot_matrix_registration_bot_bot_access_token`. Please also remove the following deprecated settings
Thanks to [Shreyas Ajjarapu](https://github.com/shreyasajj)'s efforts, the playbook now supports bridging to [Google Messages](https://messages.google.com/) via the [mautrix-gmessages](https://github.com/mautrix/gmessages) bridge. See our [Setting up Mautrix Google Messages bridging](docs/configuring-playbook-bridge-mautrix-gmessages.md) documentation page for getting started.
Thanks to [Michael Hollister](https://github.com/Michael-Hollister) from [FUTO](https://www.futo.org/), the creators of the [Circles app](https://circu.li/), the playbook can now set up [matrix-media-repo](https://github.com/turt2live/matrix-media-repo) - an alternative way to store homeserver media files, powered by a homeserver-independent implementation which supports S3 storage, IPFS, deduplication and other advanced features.
To learn more see our [Storing Matrix media files using matrix-media-repo](docs/configuring-playbook-matrix-media-repo.md) documentation page.
## Enabling `forget_rooms_on_leave` by default for Synapse
With the [Synapse v1.84.0 update](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/2698), we've also **changed the default value** of the `forget_rooms_on_leave` setting of Synapse to a value of `true`.
This way, **when you leave a room, Synapse will now forget it automatically**.
The upstream Synapse default is `false` (disabled), so that you must forget rooms manually after leaving.
**We go against the upstream default** ([somewhat controversially](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/2700)) in an effort to make Synapse leaner and potentially do what we believe most users would expect their homeserver to be doing.
If you'd like to go back to the old behavior, add the following to your configuration: `matrix_synapse_forget_rooms_on_leave: false`
**TLDR**: the `matrix-jitsi` role is now included from the [ansible-role-jitsi](https://github.com/mother-of-all-self-hosting/ansible-role-jitsi) repository, part of the [MASH playbook](https://github.com/mother-of-all-self-hosting/mash-playbook). Some variables have been renamed. All functionality remains intact.
The `matrix-jitsi` role has been relocated in its own repository, part of the [MASH playbook](https://github.com/mother-of-all-self-hosting/mash-playbook) project - an Ansible playbook for self-hosting [a growing list of FOSS software](https://github.com/mother-of-all-self-hosting/mash-playbook/blob/main/docs/supported-services.md). If hosting a Jitsi stack on the Matrix server itself did not stand right with you or you always wanted to host most stuff, you can now use this new playbook to do so.
As part of the extraction process of this role out of the Matrix playbook, a few other things improved:
- **native Traefik support** has been added
- **support for hosting under a subpath** has been added, although it suffers from a few minor issues listed [here](https://github.com/mother-of-all-self-hosting/mash-playbook/blob/main/docs/services/jitsi.md#url)
ntfy provides a web app, which is now disabled by default, because it may be unknown to and unused by most users of this playbook. You can enable it by setting `ntfy_web_root: "app"` (see [ntfy documentation](docs/configuring-playbook-ntfy.md)).
This change was already applied a while before this entry, but as some users were reporting the missing web app, this entry was added (see [#2529](https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/2529)).
## The matrix-prometheus role lives independently now
**TLDR**: the `matrix-prometheus` role is now included from the [ansible-role-prometheus](https://github.com/mother-of-all-self-hosting/ansible-role-prometheus) repository, part of the [MASH playbook](https://github.com/mother-of-all-self-hosting/mash-playbook). Some variables have been renamed. All functionality remains intact.
The `matrix-prometheus` role has been relocated in its own repository, part of the [MASH playbook](https://github.com/mother-of-all-self-hosting/mash-playbook) project - an Ansible playbook for self-hosting [a growing list of FOSS software](https://github.com/mother-of-all-self-hosting/mash-playbook/blob/main/docs/supported-services.md). If hosting a Prometheus stack on the Matrix server itself did not stand right with you or you always wanted to host most stuff, you can now use this new playbook to do so.
Extracting the Prometheus role out of this Matrix playbook required huge internal refactoring to the way the Prometheus configuration (scraping jobs) is generated. If you notice any breakage after upgrading, let us know.
Thanks to [Aine](https://gitlab.com/etke.cc) of [etke.cc](https://etke.cc/), the playbook can now set up [rust-synapse-compress-state](https://github.com/matrix-org/rust-synapse-compress-state)'s `synapse_auto_compressor` tool to run periodically.
If enabled, `synapse_auto_compressor` runs on a schedule and compresses your Synapse database's `state_groups` table. It was possible to run `rust-synapse-compress-state` manually via the playbook even before - see [Compressing state with rust-synapse-compress-state](docs/maintenance-synapse.md#compressing-state-with-rust-synapse-compress-state). However, using `synapse_auto_compressor` is better, because:
- it runs on a more up-to-date version of `rust-synapse-compress-state`
- it's a set-it-and-forget-it tool that you can enable and never have to deal with manual compression anymore
This tool needs to be enabled manually, for now. In the future, we're considering enabling it by default for all Synapse installations.
See our [Setting up synapse-auto-compressor](docs/configuring-playbook-synapse-auto-compressor.md) documentation to get started.
Thanks to [Benjamin Kampmann](https://github.com/gnunicorn) for [getting it started](https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/2515), [FSG-Cat](https://github.com/FSG-Cat) for fixing it up and me ([Slavi](https://github.com/spantaleev)) for polishing it up, the playbook can now install and configure the [sliding-sync proxy](https://github.com/matrix-org/sliding-sync).
The upcoming Element X clients ([Element X iOS](https://github.com/vector-im/element-x-ios) and [Element X Android](https://github.com/vector-im/element-x-android)) require the `sliding-sync` proxy to do their job. **These clients are still in beta** (especially Element X Android, which requires manual compilation to get it working with a non-`matrix.org` homeseserver). Playbook users can now easily give these clients a try and help test them thanks to us having `sliding-sync` support.
## The matrix-etherpad role lives independently now
**TLDR**: the `matrix-etherpad` role is now included from [another repository](https://gitlab.com/etke.cc/roles/etherpad). Some variables have been renamed. All functionality remains intact.
**TLDR**: new installations will now default to Traefik as their reverse-proxy. Existing users need to explicitly choose their reverse-proxy type. [Switching to Traefik](#how-do-i-switch-my-existing-setup-to-traefik) is strongly encouraged. `matrix-nginx-proxy` may break over time and will ultimately be removed.
As mentioned 2 weeks ago in [(Backward Compatibility) Reverse-proxy configuration changes and initial Traefik support](#backward-compatibility-reverse-proxy-configuration-changes-and-initial-traefik-support), the playbook is moving to Traefik as its default SSL-terminating reverse-proxy.
Until now, we've been doing the migration gradually and keeping full backward compatibility. New installations were defaulting to `matrix-nginx-proxy` (just like before), while existing installations were allowed to remain on `matrix-nginx-proxy` as well. This makes things very difficult for us, because we need to maintain and think about lots of different setups:
- Traefik managed by the playbook
- Traefik managed by the user in another way
- another reverse-proxy on the same host (`127.0.0.1` port exposure)
- another reverse-proxy on another host (`0.0.0.0` port exposure)
-`matrix-nginx-proxy` - an `nginx` container managed by the playbook
-`nginx` webserver operated by the user, running without a container on the same server
Each change we do and each new feature that comes in needs to support all these different ways of reverse-proxying. Because `matrix-nginx-proxy` was the default and pretty much everyone was (and still is) using it, means that new PRs also come with `matrix-nginx-proxy` as their main focus and Traefik as an afterthought, which means we need to spend hours fixing up Traefik support.
We can't spend all this time maintaining so many different configurations anymore. Traefik support has been an option for 2 weeks and lots of people have already migrated their server and have tested things out. Traefik is what we use and preferentially test for.
It's time for the **next step in our migration process** to Traefik and elimination of `matrix-nginx-proxy`:
- Traefik is now the default reverse-proxy for new installations
- All existing users need to explicitly choose their reverse-proxy type by defining the `matrix_playbook_reverse_proxy_type` variable in their `vars.yml` configuration file. We strongly encourage existing users to [switch the Traefik](#how-to-switch-an-existing-setup-to-traefik), as the nginx setup is bound to become more and more broken over time until it's ultimately removed
### How do I switch my existing setup to Traefik?
**For users who are on `matrix-nginx-proxy`** (the default reverse-proxy provided by the playbook), switching to Traefik can happen with a simple configuration change. Follow this section from 2 weeks ago: [How do I explicitly switch to Traefik right now?](#how-do-i-explicitly-switch-to-traefik-right-now).
If you experience trouble:
1. Follow [How do I remain on matrix-nginx-proxy?](#how-do-i-remain-on-matrix-nginx-proxy) to bring your server back online using the old reverse-proxy
2. Ask for help in our [support channels](README.md#support)
3. Try switching to Traefik again later
**For users with a more special reverse-proxying setup** (another nginx server, Apache, Caddy, etc.), the migration may not be so smooth. Follow the [Using your own webserver](docs/configuring-playbook-own-webserver.md) guide. Ideally, your custom reverse-proxy will be configured in such a way that it **fronts the Traefik reverse-proxy** provided by the playbook. Other means of reverse-proxying are more fragile and may be deprecated in the future.
### I already use my own Traefik server. How do I plug that in?
See the [Traefik managed by the playbook](docs/configuring-playbook-own-webserver.md#traefik-managed-by-the-playbook) section.
### Why is matrix-nginx-proxy used even after switching to Traefik?
This playbook manages many different services. All these services were initially integrated with `matrix-nginx-proxy`.
While we migrate all these components to have native Traefik support, some still go through nginx internally (Traefik -> local `matrix-nginx-proxy` -> component).
As time goes on, internal reliance on `matrix-nginx-proxy` will gradually decrease until it's completely removed.
### How do I remain on matrix-nginx-proxy?
Most new work and testing targets Traefik, so remaining on nginx is **not** "the good old stable" option, but rather the "still available, but largely untested and likely to be broken very soon" option.
To proceed regardless of this warning, add `matrix_playbook_reverse_proxy_type: playbook-managed-nginx` to your configuration.
At some point in the **near** future (days, or even weeks at most), we hope to completely get rid of `matrix-nginx-proxy` (or break it enough to make it unusable), so you **will soon be forced to migrate** anyway. Plan your migration accordingly.
### How do I keep using my own other reverse-proxy?
We recommend that you follow the guide for [Fronting the integrated reverse-proxy webserver with another reverse-proxy](docs/configuring-playbook-own-webserver.md#fronting-the-integrated-reverse-proxy-webserver-with-another-reverse-proxy).
Thanks to [Benjamin Kampmann](https://github.com/gnunicorn), the playbook can now install and configure the [Rageshake](https://github.com/matrix-org/rageshake) bug report server.
Additional details are available in [Setting up Rageshake](docs/configuring-playbook-rageshake.md).
The playbook can now help you customize Synapse's templates.
Additional details are available in the [Customizing templates](docs/configuring-playbook-synapse.md#customizing-templates) section of our Synapse documentation.
**TLDR**: the `matrix-redis` role is now included from another repository. Some variables have been renamed. All functionality remains intact.
The `matrix-redis` role (which configures [Redis](https://redis.io/)) has been extracted from the playbook and now lives in its [own repository](https://gitlab.com/etke.cc/roles/redis). This makes it possible to easily use it in other Ansible playbooks.
You need to **update your roles** (`just roles` or `make roles`) regardless of whether you're enabling Ntfy or not. If you're making use of Ntfy via this playbook, you will need to update variable references in your `vars.yml` file (`matrix_redis_` -> `redis_`).
**TLDR**: the `matrix-ntfy` role is now included from another repository. Some variables have been renamed. All functionality remains intact.
The `matrix-ntfy` role (which configures [Ntfy](https://ntfy.sh/)) has been extracted from the playbook and now lives in its [own repository](https://gitlab.com/etke.cc/roles/ntfy). This makes it possible to easily use it in other Ansible playbooks.
You need to **update your roles** (`just roles` or `make roles`) regardless of whether you're enabling Ntfy or not. If you're making use of Ntfy via this playbook, you will need to update variable references in your `vars.yml` file (`matrix_ntfy_` -> `ntfy_`).
## The matrix-grafana role lives independently now
**TLDR**: the `matrix-grafana` role is now included from another repository. Some variables have been renamed. All functionality remains intact.
The `matrix-grafana` role (which configures [Grafana](docs/configuring-playbook-prometheus-grafana.md)) has been extracted from the playbook and now lives in its [own repository](https://gitlab.com/etke.cc/roles/grafana). This makes it possible to easily use it in other Ansible playbooks.
You need to **update your roles** (`just roles` or `make roles`) regardless of whether you're enabling Grafana or not. If you're making use of Grafana via this playbook, you will need to update variable references in your `vars.yml` file (`matrix_grafana_` -> `grafana_`).
Thanks to [moan0s](https://github.com/moan0s), the `matrix-backup-borg` role (which configures [Borg backups](docs/configuring-playbook-backup-borg.md)) has been extracted from the playbook and now lives in its [own repository](https://gitlab.com/etke.cc/roles/backup_borg). This makes it possible to easily use it in other Ansible playbooks and will become part of [nextcloud-docker-ansible-deploy](https://github.com/spantaleev/nextcloud-docker-ansible-deploy) soon.
You need to **update your roles** (`just roles` or `make roles`) regardless of whether you're enabling Borg backup functionality or not. If you're making use of Borg backups via this playbook, you will need to update variable references in your `vars.yml` file (`matrix_backup_borg_` -> `backup_borg_`).
- there's a new `matrix_playbook_reverse_proxy_type` variable (see [roles/custom/matrix-base/defaults/main.yml](roles/custom/matrix-base/defaults/main.yml)), which lets you tell the playbook what reverse-proxy setup you'd like to have. This makes it easier for people who want to do reverse-proxying in other ways.
- the default reverse-proxy (`matrix_playbook_reverse_proxy_type`) is still `playbook-managed-nginx` (via `matrix-nginx-proxy`), for now. **Existing `matrix-nginx-proxy` users should not observe any changes** and can stay on this for now.
- **Users who use their [own other webserver](docs/configuring-playbook-own-webserver.md) (e.g. Apache, etc.) need to change** `matrix_playbook_reverse_proxy_type` to something like `other-on-same-host`, `other-on-another-host` or `other-nginx-non-container`
- we now have **optional [Traefik](https://traefik.io/) support**, so you could easily host Matrix and other Traefik-native services in containers on the same server. Traefik support is still experimental (albeit, good enough) and will improve over time. It does work, but certain esoteric features may not be there yet.
- **Traefik will become the default reverse-proxy in the near future**. `matrix-nginx-proxy` will either remain as an option, or be completely removed to simplify the playbook
### Motivation for redoing our reverse-proxy setup
The playbook has supported various reverse-proxy setups for a long time.
We have various configuration variables (`matrix_nginx_proxy_enabled`, various `_host_bind_port` variables, etc.) which allow the playbook to adapt to these different setups. The whole situation was messy though - hard to figure out and with lots of variables to toggle to make things work as you'd expect - huge **operational complexity**.
We love containers, proven by the fact that **everything** that this playbook manages runs in a container. Yet, we weren't allowing people to easily host other web-exposed containers alongside Matrix services on the same server. We were using `matrix-nginx-proxy` (our integrated [nginx](https://nginx.org/) server), which was handling web-exposure and SSL termination for our own services, but we **weren't helping you with all your other containers**.
People who were **using `matrix-nginx-proxy`** were on the happy path on which everything worked well by default (Matrix-wise), **but** could not easily run other web-exposed services on their Matrix server because `matrix-nginx-proxy` was occupying ports `80` and `443`. Other services which wanted to get web exposure either had to be plugged into `matrix-nginx-proxy` (somewhat difficult) or people had to forgo using `matrix-nginx-proxy` in favor of something else.
Of those that decided to forgo `matrix-nginx-proxy`, many were **using nginx** on the same server without a container. This was likely some ancient nginx version, depending on your choice of distro. The Matrix playbook was trying to be helpful and even with `matrix_nginx_proxy_enabled: false` was still generating nginx configuration in `/matrix/nginx-proxy/conf.d`. Those configuration files were adapted for inclusion into an nginx server running locally. Disabling the `matrix-nginx-proxy` role like this, yet still having it produce files is a bit disgusting, but it's what we've had since the early beginnings of this playbook.
Others still, wanted to run Matrix locally (no SSL certificates), regardless of which web server technology this relied on, and then **reverse-proxy from another machine on the network** which was doing SSL termination. These people were:
- *either* relying on `matrix_nginx_proxy_enabled: false` as well, combined with exposing services manually (setting `_bind_port` variables)
- *or* better yet, they were keeping `matrix-nginx-proxy` enabled, but in `http`-only mode (no SSL certificate retrieval).
Despite this operational complexity, things worked and were reasonably flexible to adapt to all these situations.
When using `matrix-nginx-proxy` as is, we still had another problem - one of **internal playbook complexity**. Too many services need to be web-exposed (port 80/443, SSL certificates). Because of this, they all had to integrate with the `matrix-nginx-proxy` role. Tens of different roles explicitly integrating with `matrix-nginx-proxy` is not what we call clean. The `matrix-nginx-proxy` role contains variables for many of these roles (yikes). Other roles were more decoupled from it and were injecting configuration into `matrix-nginx-proxy` at runtime - see all the `inject_into_nginx_proxy.yml` task files in this playbook (more decoupled, but still.. yikes).
The next problem is one of **efficiency, interoperability and cost-saving**. We're working on other playbooks:
- [vaultwarden-docker-ansible-deploy](https://github.com/spantaleev/vaultwarden-docker-ansible-deploy) for hosting the [Vaultwarden](https://github.com/dani-garcia/vaultwarden) server - an alternative implementation of the [Bitwarden](https://bitwarden.com/) password manager
- [gitea-docker-ansible-deploy](https://github.com/spantaleev/gitea-docker-ansible-deploy) - for hosting the [Gitea](https://gitea.io/) git source code hosting service
- [nextcloud-docker-ansible-deploy](https://github.com/spantaleev/nextcloud-docker-ansible-deploy) - for hosting the [Nextcloud](https://nextcloud.com/) groupware platform
We'd love for users to be able to **seamlessly use all these playbooks (and others, even) against a single server**. We don't want `matrix-nginx-proxy` to have a monopoly on port `80`/`443` and make it hard for other services to join in on the party. Such a thing forces people into running multiple servers (one for each service), which does provide nice security benefits, but is costly and ineffiecient. We'd like to make self-hosting these services cheap and easy.
These other playbooks have been using [Traefik](https://traefik.io/) as their default reverse-proxy for a long time. They can all coexist nicely together (as an example, see the [Interoperability](https://github.com/spantaleev/nextcloud-docker-ansible-deploy/blob/master/docs/configuring-playbook-interoperability.md) documentation for the [Nextcloud playbook](https://github.com/spantaleev/nextcloud-docker-ansible-deploy)). Now that this playbook is gaining Traefik support, it will be able to interoperate with them. If you're going this way, make sure to have the Matrix playbook install Traefik and have the others use `*_reverse_proxy_type: other-traefik-container`.
Finally, at [etke.cc - a managed Matrix server hosting service](https://etke.cc) (built on top of this playbook, and coincidentally [turning 2 years old today](https://etke.cc/news/upsyw4ykbtgmwhz8k7ukldx0zbbfq-fh0iqi3llixi0/) 🎉), we're allowing people to host some additional services besides Matrix components. Exposing these services to the web requires ugly hacks and configuration files being dropped into `/matrix/nginx-proxy/conf.d`. We believe that everything should run in independent containers and be exposed to the web via a Traefik server, without a huge Ansible role like `matrix-nginx-proxy` that everything else needs to integrate with.
### How do these changes fix all these problems?
The new `matrix_playbook_reverse_proxy_type` lets you easily specify your preferred reverse-proxy type, including `other-on-same-host`, `other-on-another-host` and `none`, so people who'd like to reverse-proxy with their own web server have more options now.
Using Traefik greatly simplifies things, so going forward we'll have a simpler and easier to maintain playbook, which is also interoperable with other services.
Traefik is a web server, which has been specifically **designed for reverse-proxying to services running in containers**. It's ideal for usage in an Ansible playbook which runs everything in containers.
