Monitoring and alerts
The server hosts its own monitoring: a status page, resource graphs and push alerts when a service goes down, a scheduled job goes quiet, or disk, CPU or memory run hot. Alerts reach you by email, through the server’s outgoing mail, and anyone else through channels you add in Healthchecks.
How it works
Section titled “How it works”Gatus (gatus.yourdomain.com) is the status page. Infrastructure services are probed every 60 to 120 seconds, and each routed, running app gets one public check every 120 seconds, regenerated every 10 minutes. An app’s own vps.health.path and vps.health.expect labels are used when present; otherwise the check reads the app’s front page: an app behind sign-in is healthy when it sends the visitor to sign in, and a public app when its front page answers with a page or a redirect rather than an error (see Configure an app for Catena). An alert fires after 3 consecutive failures and resolves after 2 successes, so a short outage takes a few minutes to show.
Healthchecks (healthchecks.yourdomain.com) receives Gatus alerts, one check per service, created on the first failure. It also receives a ping from each scheduled job (backup, offsite copy, nightly maintenance, restart probe, off-site heartbeat); a missed ping is the alarm. Alerts reach the admin email by default, through the mail service chosen in Settings > Outgoing mail; you add more notification channels (email, Slack and many others) in Healthchecks’ own interface.
Beszel (beszel.yourdomain.com) graphs server resources through an agent on the host, with the history and the usage of each container; the System tab’s CPU and RAM gauges are read from the server itself and do not depend on it. Alert rules are created once for the server: status (agent silent) after 5 minutes, CPU and memory at 90 percent and disk at 85 percent, each sustained for 10 minutes. Alerts go through a Healthchecks bridge, so they reach the admin email and the other channels like any other alert. Rules are never overwritten after creation, so later edits in Beszel stay.
Off-site heartbeat. The Healthchecks on the server cannot report an outage that takes the whole server down. With Catena Pro, the off-site heartbeat calls a check at an outside monitoring service on the schedule you set in Schedules; you save its address in Settings > Alerts and missed-job reporting. When the calls stop, the outside service raises the alarm, and a call that names a silent monitor reports the monitor that stopped.
Other signals. The panel shows tiles for monitoring status and exposed ports, a banner for missing backups and pending restarts, and a System page with active alerts and jobs that have not yet reported. Antivirus watch and mail delivery canaries run as timers.
What each edition adds
Section titled “What each edition adds”Gatus, Healthchecks, Beszel, the status page and on-host alerts work on every edition, and so do the security alerts: the vulnerability watch emails you about an exploited vulnerability, or a critical one with a fixed version (see Security alerts). Catena Pro adds the off-site heartbeat and Catena Business adds monitoring-as-a-service, described in the edition comparison.
Limits
Section titled “Limits”- On-server monitors cannot report an outage of the server itself, which is why the off-site heartbeat exists.
- Alerting needs a channel: with No outgoing mail, nothing is sent until you add a channel in Healthchecks.
- Alert latency is the probe interval times the 3-failure threshold.