Escalation

A monitor’s own subscriptions notify once, right when it triggers. Escalation adds a chain of follow-up notifications for a triggered monitor that nobody’s acknowledged yet — so an incident that’s being ignored gets louder instead of going quiet after the first alert.

The step chain

An escalation chain is an ordered list of steps, each with:

• Delay — minutes from when this step becomes due.
• Destination — which notification destination this step notifies.

Step 1’s delay counts from when the monitor first triggered the current incident. Every step after that counts from when the previous step fired, not from the incident’s start — so a chain of “5 min, then 10 min, then 15 min” notifies at 5 minutes, 15 minutes, and 30 minutes after the trigger, each gap measured from the step before it.

A step only fires if the monitor is still triggered, still unacknowledged, and neither muted nor snoozed when it comes due. If the monitor clears to ok, gets acknowledged, is muted, or is snoozed before a step’s delay elapses, that step never fires.

Acknowledging halts the chain

Acknowledging a triggered monitor stops any further escalation steps from firing — it’s the “I’ve got this” signal that tells arcodash to stop notifying the next destination in line. It doesn’t fire the steps already sent, and it doesn’t mute the monitor’s normal state-change notifications.

Going from ok (or unknown) into triggered starts a fresh incident: any earlier acknowledgment is cleared and escalation restarts from its first step. If the monitor triggers again later, its escalation chain runs again from step one.

Replacing the chain

PUT /api/monitors/:id/escalation replaces the entire chain in one call — send the full ordered list of steps you want, not just the one you’re adding or changing. Positions 1..N are assigned from the array’s order, so reordering is just resubmitting the array in the order you want. PUT an empty array to clear the chain entirely.

PUT /api/monitors/42/escalation
[
  { "delay_minutes": 5, "destination_id": 4 },
  { "delay_minutes": 10, "destination_id": 7 }
]

GET /api/monitors/:id/escalation returns the current chain in the same shape.

Progress indicator

A monitor’s own page shows where it is in its chain: “no steps notified yet, N configured” before the first step fires, or “step X of N notified” once at least one has. This resets to zero the same moment escalation restarts — a fresh incident, a fresh count.

Where to go next

• Alerts overview — the trigger condition and state machine escalation builds on top of.
• Managing monitor incidents — acknowledge, mute, snooze, and delivery history.
• Notification destinations — what each escalation step actually notifies.