Skip to content

Severity tiers

A rule’s severity tiers are stricter levels of the same rule. The rule fires at its own severity; when the value or the count crosses a tier, the same alert moves up to that tier’s severity.

Rule Tiers
Response time > 800 ms, medium > 1200 ms high, > 2000 ms critical
2 of the last 5 runs failed, medium 4 of 5 critical
  • It stays one alert. There is no second alert for the higher tier, and no second ticket or incident.
  • Rising a tier notifies. The escalated notice goes to the channels the alert already reached, and to any channel the new severity routes to. So a Jira-only medium alert can page PagerDuty once it turns critical.
  • Falling back is quiet. The alert’s severity goes down and its timeline records it, but nobody is notified.
  • Resolving works as usual, once the rule’s base condition is clear.

Severities, from highest: critical, high, medium, low, info.

In the rule form, under Severity tiers, add a level with its severity and its stricter limit (a threshold for threshold rules, a count for count rules).

  • A rule has at most 4 tiers.
  • Each tier must be stricter than the one before it, and more severe.
  • A count tier can’t ask for more runs than the window holds.
  • Threshold tiers work with >, >=, < and <=, not == or !=.

In the API, tiers are the rule’s escalations:

{ "name": "Latency", "severity": "medium",
"condition": { "type": "threshold", "value": "durationMs", "comparator": ">", "threshold": 800,
"escalations": [ { "severity": "high", "threshold": 1200 }, { "severity": "critical", "threshold": 2000 } ] } }

See the alert rule reference for the count form.