Escalation Policies

Policy Structure

Escalation policies define how incidents are escalated when they're not acknowledged or resolved within specified timeframes. Each policy has multiple layers, each with a target (user, group, or on-call schedule), notification method, and timeout. If the first layer doesn't respond, the escalation advances to the next layer automatically.

On-Call Schedules

On-call schedules are configured in Workflows → Escalation → On-Call Schedules. Schedules support weekly rotations, override slots, and multiple simultaneous on-call assignees (primary + secondary). Schedule data is consumed by the escalation engine to determine who is currently on-call when an escalation fires.

{
  "on_call_schedule": {
    "name": "Platform Engineering On-Call",
    "timezone": "America/New_York",
    "rotation": {
      "type": "weekly",
      "start_day": "Monday",
      "start_time": "09:00",
      "participants": [
        {"user_id": "usr_alice_johnson", "weeks": [1, 3, 5]},
        {"user_id": "usr_bob_chen", "weeks": [2, 4]},
        {"user_id": "usr_carol_davis", "weeks": [6, 7, 8]}
      ]
    }
  }
}

Escalation Triggers

Escalations can be triggered by: unacknowledged P1 incident (default: 15 minutes), SLA breach, manual escalation by agent, or automatic escalation by a workflow node. Triggers are evaluated every 60 seconds by the escalation timer job running within the StackFlowAPI Lambda.

Notification Methods

Testing Escalation

Test escalation policies without creating real incidents using the policy test tool at Workflows → Escalation → Test Policy. Provide a test policy ID and a test incident (or use a dummy incident) and the test runner simulates the escalation timeline, showing which user would be notified at each stage without actually sending notifications.