Alerts
MonoPbx watches itself and raises named alerts. Installs with Prometheus use the shipped rule files. Installs without it use the alert centre in the console. Both raise the same alerts under the same names.
The alert centre
Open Admin, Alerts (needs system.settings:read). The PBX checks its own metrics every 30 seconds.
- An alert is raised when its condition has held for the rule's waiting time, and resolved when the condition clears.
- Acknowledge records that you saw it. It does not silence it; the alert keeps firing until the problem is gone.
- Resolved alerts are kept 90 days.
- Alerts describe the host and its processes. They never carry a number, a person or a company.
Where alerts are sent
On the same page, Where alerts are sent (needs system.settings:write):
| Setting | Meaning |
|---|---|
| Send notifications | Off keeps the alert centre but sends nothing. |
| Minimum severity | Info, warning (default) or critical. |
| Also when an alert clears | Sends a second message when the alert resolves (default on). |
| E-mail recipients | Up to 10 addresses. Mail goes through the install's mail settings, not a company's. The page tells you when none are set. |
| Webhook | A public https address. Deliveries are signed like other webhooks, so any Standard Webhooks library can verify them. Internal addresses are refused. |
The test button sends a harmless test alert to every channel and shows each result.
What is watched
| Alert | Severity | Meaning |
|---|---|---|
TrunkDown | critical | A trunk fails its health check for 5 minutes. |
RegistrationsDropped | critical | Registered phones fell by more than 30 % in 5 minutes (from at least 10). |
HighCallFailureRatio | warning | More than 30 % of recent calls failed. |
SipBanSpike | warning | More than 20 SIP sources banned in 10 minutes. |
SipAccountLocked, SipCompanyUnderAttack | warning, critical | Failed authentication, see SIP edge protection. |
RtpengineNodeDown, MediaServiceDown | critical | The media path is down; new calls fail. |
PushWakeFailuresHigh, PushRelayUnreachable | warning, critical | Mobile wake-up pushes fail. |
PushEntitlementGrace | warning | The push entitlement was revoked; a 30-day grace period runs. |
SmppBindDown | critical | An SMS provider session is not bound. |
AiProviderErrors | warning | More than half the AI requests fail. |
DiskSpaceLow | warning | Under 10 % free on a watched file system. |
CertExpiringSoon, CertExpiryCritical, CertRenewalFailing, CertIssuanceFailed | warning, critical | Certificate problems, see Certificates. |
StirCertExpiringSoon, StirCertExpiryCritical | warning, critical | A STIR/SHAKEN signing certificate expires soon. |
SiemExporterDown | warning | The log export cannot reach its collector. |
ClickHouseSpillGrowing | warning | Analytics storage is not taking records. |
A condition with no data counts as nothing wrong. Alerts that need volume (registrations, call failures) stay quiet on a quiet install.
With Prometheus
Load deploy/compose/prometheus/rules/*.yml into your Prometheus. The optional observability bundle already does. A starting Alertmanager configuration is included: critical alerts page at once and repeat every 2 hours, warnings group for 10 minutes and repeat daily. Alertmanager itself is not bundled. See error codes for alerts.