Cookbook

Task-first recipes: what you’ll get, what you need first, the steps, and how to check it worked. Each one links back to the fuller reference pages for the pieces it uses.

Get set up

Recipe	You’ll get
Proxmox to Telegram in five minutes	A Proxmox VE or Backup Server source posting straight to a Telegram chat.

Watch a job or a host

Recipe	You’ll get
Watch a nightly backup job	An alert when a cron’d backup script (restic, borg, rsync, or your own) doesn’t start, fails, or never checks in.
Know when a host goes dark	A liveness heartbeat that fires when a machine stops checking in — plus the honest limit of doing this from the machine itself.
Alert on failed systemd units	Any unit’s failure reported the moment systemd notices it.
ZFS and disk health alerts	Pool and disk events from ZED and smartd turned into InfraInbox incidents, deduplicated per pool or disk.
Failed CI jobs (GitHub Actions)	A workflow step that reports a failed run, and resolves the incident when a later run succeeds.

Shape your alerts

Recipe	You’ll get
Only critical alerts at night	Notices below your chosen severity held back overnight — still visible in the inbox, just not pushed to you.
Send one app’s alerts to its own channel	A routing rule that sends a specific source or label to its own destination instead of the default route.
Silence alerts during maintenance	A time-boxed silence so planned work doesn’t page anyone.
Tame a flapping monitor	Fewer notifications from a check that keeps flipping between up and down.

Build your own integration

Recipe	You’ll get
Verify a signed webhook	Working Python and Go/Node snippets that check InfraInbox’s outbound webhook signature before trusting a payload.