Escalation policies for on-call alertsEscalate to whoever is on call

Name an escalation policy on the send. If nobody acknowledges, each step waits and then alerts the next people. The first acknowledgement stops it on every device it reached.

A step can point at a rotation. Acciti reads the rotation when the step runs, so when someone adds holiday cover the policy follows without an edit.

curl
curl https://acciti.com/v1/messages \  -H "Authorization: Bearer $ACCITI_TOKEN" \  -H "Content-Type: application/json" \  -d '{    "user": "your-user-key",    "title": "api.acme.com is down",    "message": "No response for 90 seconds",    "priority": 2,    "retry": 60,    "expire": 1800,    "escalation": "production"  }'

A policy name your team doesn't have is refused with unknown_escalation. The alert isn't sent without it, so you never believe a policy is running when it isn't.

Build policies and rotations on the dashboard. The on-call page shows how they fit together.

Planned work

Book a maintenance window on the same page before you take a host down. While it's open, alerts from the senders it covers go to the inbox marked "held": "maintenance" and start no escalation. Other senders are unaffected, and everything held stays in the inbox.