dendrite/docs/INSTALL.md

318 lines
10 KiB
Markdown
Raw Permalink Normal View History

# Installing Dendrite
Dendrite can be run in one of two configurations:
* **Polylith mode**: A cluster of individual components, dealing with different
aspects of the Matrix protocol (see [WIRING.md](WIRING-Current.md)). Components communicate
with each other using internal HTTP APIs and [Apache Kafka](https://kafka.apache.org).
This will almost certainly be the preferred model for large-scale deployments.
* **Monolith mode**: All components run in the same process. In this mode,
Kafka is completely optional and can instead be replaced with an in-process
lightweight implementation called [Naffka](https://github.com/matrix-org/naffka). This
will usually be the preferred model for low-volume, low-user or experimental deployments.
For most deployments, it is **recommended to run in monolith mode with PostgreSQL databases**.
Regardless of whether you are running in polylith or monolith mode, each Dendrite component that
requires storage has its own database. Both Postgres and SQLite are supported and can be
mixed-and-matched across components as needed in the configuration file.
Be advised that Dendrite is still in development and it's not recommended for
use in production environments just yet!
## Requirements
Dendrite requires:
* Go 1.13 or higher
* Postgres 9.6 or higher (if using Postgres databases, not needed for SQLite)
If you want to run a polylith deployment, you also need:
* Apache Kafka 0.10.2+
Please note that Kafka is **not required** for a monolith deployment.
## Building Dendrite
Start by cloning the code:
```bash
git clone https://github.com/matrix-org/dendrite
cd dendrite
```
Then build it:
```bash
2019-05-21 20:56:55 +00:00
./build.sh
```
## Install Kafka (polylith only)
Install and start Kafka (c.f. [scripts/install-local-kafka.sh](scripts/install-local-kafka.sh)):
```bash
KAFKA_URL=http://archive.apache.org/dist/kafka/2.1.0/kafka_2.11-2.1.0.tgz
# Only download the kafka if it isn't already downloaded.
test -f kafka.tgz || wget $KAFKA_URL -O kafka.tgz
# Unpack the kafka over the top of any existing installation
mkdir -p kafka && tar xzf kafka.tgz -C kafka --strip-components 1
# Start the zookeeper running in the background.
# By default the zookeeper listens on localhost:2181
kafka/bin/zookeeper-server-start.sh -daemon kafka/config/zookeeper.properties
# Start the kafka server running in the background.
# By default the kafka listens on localhost:9092
kafka/bin/kafka-server-start.sh -daemon kafka/config/server.properties
```
On macOS, you can use [Homebrew](https://brew.sh/) for easier setup of Kafka:
```bash
brew install kafka
brew services start zookeeper
brew services start kafka
```
## Configuration
### PostgreSQL database setup
Assuming that PostgreSQL 9.6 (or later) is installed:
* Create role, choosing a new password when prompted:
```bash
sudo -u postgres createuser -P dendrite
```
2020-12-14 09:42:27 +00:00
At this point you have a choice on whether to run all of the Dendrite
components from a single database, or for each component to have its
own database. For most deployments, running from a single database will
be sufficient, although you may wish to separate them if you plan to
split out the databases across multiple machines in the future.
On macOS, omit `sudo -u postgres` from the below commands.
* If you want to run all Dendrite components from a single database:
```bash
sudo -u postgres createdb -O dendrite dendrite
```
... in which case your connection string will look like `postgres://user:pass@database/dendrite`.
* If you want to run each Dendrite component with its own database:
```bash
for i in mediaapi syncapi roomserver signingkeyserver federationsender appservice keyserver userapi_accounts userapi_devices naffka; do
sudo -u postgres createdb -O dendrite dendrite_$i
done
```
2020-12-14 09:42:27 +00:00
... in which case your connection string will look like `postgres://user:pass@database/dendrite_componentname`.
### SQLite database setup
**WARNING:** SQLite is suitable for small experimental deployments only and should not be used in production - use PostgreSQL instead for any user-facing federating installation!
Dendrite can use the built-in SQLite database engine for small setups.
The SQLite databases do not need to be pre-built - Dendrite will
create them automatically at startup.
### Server key generation
Each Dendrite installation requires:
2020-12-14 09:42:27 +00:00
* A unique Matrix signing private key
* A valid and trusted TLS certificate and private key
To generate a Matrix signing private key:
```bash
./bin/generate-keys --private-key matrix_key.pem
```
2020-12-14 09:42:27 +00:00
**WARNING:** Make sure take a safe backup of this key! You will likely need it if you want to reinstall Dendrite, or
any other Matrix homeserver, on the same domain name in the future. If you lose this key, you may have trouble joining
federated rooms.
For testing, you can generate a self-signed certificate and key, although this will not work for public federation:
```bash
./bin/generate-keys --tls-cert server.crt --tls-key server.key
```
2020-12-14 09:42:27 +00:00
If you have server keys from an older Synapse instance,
[convert them](serverkeyformat.md#converting-synapse-keys) to Dendrite's PEM
format and configure them as `old_private_keys` in your config.
### Configuration file
Create config file, based on `dendrite-config.yaml`. Call it `dendrite.yaml`. Things that will need editing include *at least*:
* The `server_name` entry to reflect the hostname of your Dendrite server
* The `database` lines with an updated connection string based on your
2020-10-20 10:22:46 +00:00
desired setup, e.g. replacing `database` with the name of the database:
2020-12-14 09:42:27 +00:00
* For Postgres: `postgres://dendrite:password@localhost/database`, e.g.
* `postgres://dendrite:password@localhost/dendrite_userapi_account` to connect to PostgreSQL with SSL/TLS
* `postgres://dendrite:password@localhost/dendrite_userapi_account?sslmode=disable` to connect to PostgreSQL without SSL/TLS
* For SQLite on disk: `file:component.db` or `file:///path/to/component.db`, e.g. `file:userapi_account.db`
* Postgres and SQLite can be mixed and matched on different components as desired.
* The `use_naffka` option if using Naffka in a monolith deployment
There are other options which may be useful so review them all. In particular,
if you are trying to federate from your Dendrite instance into public rooms
then configuring `key_perspectives` (like `matrix.org` in the sample) can
help to improve reliability considerably by allowing your homeserver to fetch
public keys for dead homeservers from somewhere else.
2020-10-20 10:22:46 +00:00
**WARNING:** Dendrite supports running all components from the same database in
PostgreSQL mode, but this is **NOT** a supported configuration with SQLite. When
2020-10-20 10:22:46 +00:00
using SQLite, all components **MUST** use their own database file.
## Starting a monolith server
It is possible to use Naffka as an in-process replacement to Kafka when using
the monolith server. To do this, set `use_naffka: true` in your `dendrite.yaml`
configuration and uncomment the relevant Naffka line in the `database` section.
Be sure to update the database username and password if needed.
The monolith server can be started as shown below. By default it listens for
HTTP connections on port 8008, so you can configure your Matrix client to use
`http://servername:8008` as the server:
```bash
./bin/dendrite-monolith-server
```
If you set `--tls-cert` and `--tls-key` as shown below, it will also listen
for HTTPS connections on port 8448:
```bash
./bin/dendrite-monolith-server --tls-cert=server.crt --tls-key=server.key
```
## Starting a polylith deployment
The following contains scripts which will run all the required processes in order to point a Matrix client at Dendrite.
2020-10-20 10:22:46 +00:00
### nginx (or other reverse proxy)
2020-10-20 10:22:46 +00:00
This is what your clients and federated hosts will talk to. It must forward
requests onto the correct API server based on URL:
2020-10-20 10:22:46 +00:00
* `/_matrix/client` to the client API server
* `/_matrix/federation` to the federation API server
* `/_matrix/key` to the federation API server
* `/_matrix/media` to the media API server
2020-10-20 10:22:46 +00:00
See `docs/nginx/polylith-sample.conf` for a sample configuration.
### Client API server
This is what implements CS API endpoints. Clients talk to this via the proxy in
order to send messages, create and join rooms, etc.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml clientapi
```
### Sync server
This is what implements `/sync` requests. Clients talk to this via the proxy
in order to receive messages.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml syncapi
```
### Media server
This implements `/media` requests. Clients talk to this via the proxy in
order to upload and retrieve media.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml mediaapi
```
### Federation API server
This implements the federation API. Servers talk to this via the proxy in
order to send transactions. This is only required if you want to support
federation.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml federationapi
```
### Internal components
This refers to components that are not directly spoken to by clients. They are only
contacted by other components. This includes the following components.
#### Room server
This is what implements the room DAG. Clients do not talk to this.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml roomserver
```
#### Federation sender
This sends events from our users to other servers. This is only required if
you want to support federation.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml federationsender
```
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 16:34:59 +00:00
#### Appservice server
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 16:34:59 +00:00
This sends events from the network to [application
services](https://matrix.org/docs/spec/application_service/unstable.html)
running locally. This is only required if you want to support running
application services on your homeserver.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml appservice
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 16:34:59 +00:00
```
#### Key server
This manages end-to-end encryption keys for users.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml keyserver
```
#### Signing key server
This manages signing keys for servers.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml signingkeyserver
```
#### EDU server
This manages processing EDUs such as typing, send-to-device events and presence. Clients do not talk to
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml eduserver
```
#### User server
This manages user accounts, device access tokens and user account data,
amongst other things.
```bash
./bin/dendrite-polylith-multi --config=dendrite.yaml userapi
```