0
0
Fork 0
mirror of https://github.com/matrix-org/dendrite synced 2024-11-10 03:41:16 +01:00
dendrite/docs/installation/1_planning.md

110 lines
5.6 KiB
Markdown
Raw Normal View History

---
title: Planning your installation
parent: Installation
nav_order: 1
permalink: /installation/planning
---
# Planning your installation
## Modes
Dendrite consists of several components, each responsible for a different aspect of the Matrix protocol.
Users can run Dendrite in one of two modes which dictate how these components are executed and communicate.
* **Monolith mode** runs all components in a single process. Components communicate through an internal NATS
server with generally low overhead. This mode dramatically simplifies deployment complexity and offers the
best balance between performance and resource usage for low-to-mid volume deployments.
* **Polylith mode** runs all components in isolated processes. Components communicate through an external NATS
server and HTTP APIs, which incur considerable overhead. While this mode allows for more granular control of
resources dedicated toward individual processes, given the additional communications overhead, it is only
necessary for very large deployments.
Given our current state of development, **we recommend monolith mode** for all deployments.
## Databases
Dendrite can run with either a PostgreSQL or a SQLite backend. There are considerable tradeoffs
to consider:
* **PostgreSQL**: Needs to run separately to Dendrite, needs to be installed and configured separately
and and will use more resources over all, but will be **considerably faster** than SQLite. PostgreSQL
has much better write concurrency which will allow Dendrite to process more tasks in parallel. This
will be necessary for federated deployments to perform adequately.
* **SQLite**: Built into Dendrite, therefore no separate database engine is necessary and is quite
a bit easier to set up, but will be much slower than PostgreSQL in most cases. SQLite only allows a
single writer on a database at a given time, which will significantly restrict Dendrite's ability
to process multiple tasks in parallel.
At this time, we **recommend the PostgreSQL database engine** for all production deployments.
## Requirements
Dendrite will run on Linux, macOS and Windows Server. It should also run fine on variants
of BSD such as FreeBSD and OpenBSD. We have not tested Dendrite on AIX, Solaris, Plan 9 or z/OS —
your mileage may vary with these platforms.
It is difficult to state explicitly the amount of CPU, RAM or disk space that a Dendrite
installation will need, as this varies considerably based on a number of factors. In particular:
* The number of users using the server;
* The number of rooms that the server is joined to — federated rooms in particular will typically
use more resources than rooms with only local users;
* The complexity of rooms that the server is joined to — rooms with more members coming and
going will typically be of a much higher complexity.
Some tasks are more expensive than others, such as joining rooms over federation, running state
resolution or sending messages into very large federated rooms with lots of remote users. Therefore
you should plan accordingly and ensure that you have enough resources available to endure spikes
in CPU or RAM usage, as these may be considerably higher than the idle resource usage.
At an absolute minimum, Dendrite will expect 1GB RAM. For a comfortable day-to-day deployment
which can participate in federated rooms for a number of local users, be prepared to assign 2-4
CPU cores and 8GB RAM — more if your user count increases.
If you are running PostgreSQL on the same machine, allow extra headroom for this too, as the
database engine will also have CPU and RAM requirements of its own. Running too many heavy
services on the same machine may result in resource starvation and processes may end up being
killed by the operating system if they try to use too much memory.
## Dependencies
In order to install Dendrite, you will need to satisfy the following dependencies.
### Go
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
2022-07-11 15:31:31 +02:00
At this time, Dendrite supports being built with Go 1.18 or later. We do not support building
Dendrite with older versions of Go than this. If you are installing Go using a package manager,
you should check (by running `go version`) that you are using a suitable version before you start.
### PostgreSQL
If using the PostgreSQL database engine, you should install PostgreSQL 12 or later.
### NATS Server
Monolith deployments come with a built-in [NATS Server](https://github.com/nats-io/nats-server) and
therefore do not need this to be manually installed. If you are planning a monolith installation, you
do not need to do anything.
Polylith deployments, however, currently need a standalone NATS Server installation with JetStream
enabled.
To do so, follow the [NATS Server installation instructions](https://docs.nats.io/running-a-nats-service/introduction/installation) and then [start your NATS deployment](https://docs.nats.io/running-a-nats-service/introduction/running). JetStream must be enabled, either by passing the `-js` flag to `nats-server`,
or by specifying the `store_dir` option in the the `jetstream` configuration.
### Reverse proxy
A reverse proxy such as [Caddy](https://caddyserver.com), [NGINX](https://www.nginx.com) or
[HAProxy](http://www.haproxy.org) is required for polylith deployments and is useful for monolith
deployments. Configuring those is not covered in this documentation, although sample configurations
for [Caddy](https://github.com/matrix-org/dendrite/blob/main/docs/caddy) and
[NGINX](https://github.com/matrix-org/dendrite/blob/main/docs/nginx) are provided.
### Windows
Finally, if you want to build Dendrite on Windows, you will need need `gcc` in the path. The best
way to achieve this is by installing and building Dendrite under [MinGW-w64](https://www.mingw-w64.org/).