An alert per integration that fires when it stops doing anything, which is the category nobody adds.
- alert: IntegrationSilent
expr: |
sum by (integration) (
increase(integration_success_total[6h])
) == 0
for: 1h
six integrations, six alerts, and each needs a window
that matches its cadence:
the supplier dispatch twice daily → 18h
the payment webhook continuous → 1h
the accounting export monthly → not an
alert. a calendar entry.
Every alert we had was for a value being too high and none for a value not arriving, which is how an integration fails silently for eleven weeks. The monthly export is the case that shows the limit — an absence alert needs a window shorter than the interval, and some intervals are too long for an alert to be the right tool.