فارسی · Components · Home
Administration
Watching the server, backing it up off-site, and housekeeping. interzone's own Manage screen already covers a lot of this without installing anything: the Dashboard, Backups of the database, Alerts, Disk space. See the setup guide.
How to add any of these: Components.
Prometheus + Grafana
Graphs of everything over time: CPU, memory, disk, network, Synapse's activity and slowness, Postgres.
DNS: stats.example.org.
Settings: Grafana admin user (admin) and password (made up for
you; ctrl+r shows it). They're for Grafana only, not a Matrix account.
After deploy: open https://stats.example.org, sign in, and open
Dashboards: Synapse, Node Exporter (the server itself) and Postgres are
there already.
Good to know: anyone who can see these graphs can guess a lot about how your server is used. Keep the password to yourself. Playbook doc
Alertmanager receiver
Forwards alerts from a Prometheus Alertmanager into Matrix rooms. The
playbook doesn't install Alertmanager itself; this is for people who run
one. For simple alerts without any of that, use Manage › Alerts (A).
Settings:
- Bot access token: of an account made for it (see getting a bot's access token).
- Room mapping: names for rooms:
alerts: "!ROOMID:example.org". Invite the bot to each room.
After deploy: point Alertmanager's webhook at
https://matrix.example.org/matrix-alertmanager-receiver-…/alert/alerts; the
playbook doc has the full address and an example configuration.
Playbook doc
synapse-usage-exporter
Synapse only. Turns Synapse's usage statistics (users, rooms, messages per day) into Prometheus metrics, for a Grafana graph. Useful only with Prometheus + Grafana. Playbook doc
synapse-auto-compressor
Synapse only. Shrinks Synapse's biggest database tables every night at midnight, so the database grows slower. Manage › Compress Synapse's state does the same once, by hand. Playbook doc
System cleanup
Removes the Docker images and containers no service uses any more, every
day, so updates don't slowly fill the disk. Manage › Disk space does it
once, by hand. To see what it would remove, run
/matrix/system-cleanup/bin/cleanup-docker --dry-run on the server.
Playbook doc
BorgBackup
Full, encrypted, deduplicated backups of the whole server (/matrix and the
database) to another machine, every night around 4 o'clock. It keeps 7 daily,
4 weekly, 12 monthly and 2 yearly backups. interzone's Backups only cover
the database; this covers everything, media included.
Before you start: a place to store them that you reach over SSH: another server, or a storage box (Hetzner Storage Box, BorgBase, rsync.net…). Make a key for it on your computer, and add its public half there:
ssh-keygen -t ed25519 -N '' -f borg-key
cat borg-key.pub | ssh USER@HOST 'dd of=.ssh/authorized_keys oflag=append conv=notrunc'
Settings:
- Repositories:
ssh://USER@HOST/./matrix; it's created on the first backup. - Encryption passphrase: made up for you. Write it down somewhere else: without it the backups can't be read.
- SSH private key: the contents of
borg-key(not.pub): a first line with just|, then the key, every line indented by two spaces.
After deploy: to run one now, systemctl start matrix-backup-borg on the
server. Restoring is done with borg itself; see the
role's documentation.
Playbook doc