On-call & escalation

Channels decide where an alert goes. Escalation policies decide who is responsible for it, and what happens if they don’t respond.

Escalation policies

A policy is an ordered list of steps. Each step names who to notify and how long to wait before moving to the next step.

Step 1  Notify: on-call engineer (SMS + push)        Wait 5 minutes
Step 2  Notify: on-call engineer (phone call)        Wait 5 minutes
Step 3  Notify: engineering lead (phone call)        Wait 10 minutes
Step 4  Notify: #incidents (Slack), everyone (email)

Escalation stops as soon as someone acknowledges the incident, whether from the dashboard, by replying to the SMS, by pressing a key on the phone call, or from Slack. Resolution (the monitor recovering) also stops it.

Attach a policy to a monitor the same way you attach a channel. A monitor can have plain channels and a policy at the same time: channels are notified immediately and the policy runs alongside.

On-call schedules

A schedule is a rotation of team members. Reference “the on-call engineer” in a policy step and StatusCheck resolves it to whoever is on the schedule at that moment.

• Rotation: daily or weekly, with a start day and handoff time in the schedule’s time zone.
• Overrides: swap a shift for a holiday or a sick day without editing the rotation.
• Who’s on call: the current and next responders are shown on the project home so nobody has to ask.

Acknowledging and snoozing

Acknowledging an incident tells the team someone is looking and pauses escalation. It doesn’t silence recovery notifications. Snoozing pauses re-notification for a chosen period on a monitor you know is flapping while you fix it.

Team members and seats

Every project member can be on a schedule and can receive alerts on any channel. Somebody who only needs to be paged, and never opens the dashboard, can have an alert-only seat instead of a full or per responder; add the whole team.