Skip to main content
An escalation policy decides who is paged for an incident, and in which order. Its steps can name people, teams, or on-call schedules, so the page follows the rotation instead of a fixed person. Open Alerts → On-call schedules and Alerts → Escalations. Reading them requires schedules.read and alerts.read; changing schedules requires schedules.manage, and changing policies requires alerts.manage.

​
How a page reaches someone

  1. A rule fires and opens an incident. The incident uses the escalation policy of the rule.
  2. Step 1 of the policy runs: each target receives a notification event. A user target is that person, a schedule target is whoever is on call now, and a team target goes to the team as a whole.
  3. Notify policies turn each event into deliveries: the connectors, the personal order of each person, and the fallback routes.
  4. When nobody acknowledges within the Ack timeout of the step, the next step runs, until the incident is acknowledged or resolved, or the steps run out.
A silence that matches the incident stops the escalation while it lasts. With several alert nodes, each rule belongs to one node, and that node also escalates the incidents of the rule. See Role-based deployment.

​
On-call schedules

On-call schedules lists the schedules with their timezone, rotations, people, who is on call now, the next handoff, and the state. The head sums up the schedules, the people on call now, the handoffs this week, and the temporary coverage. Filter by schedule, team, owner, timezone, or state, or show Only schedules I participate in.

​
Create a schedule

  1. Select New schedule.
  2. Under Basic information, enter the Schedule name, a Description, the Timezone, such as Asia/Shanghai, and the Team.
  3. Under Rotation rules, set up each rotation:
  4. Check Next 7 days, which previews the order, the handoff times, and the people covered.
  5. Select Create schedule, or Save draft to keep it paused.
When several rotations cover the same moment, the first one in the list that has someone on call decides. Add a second rotation, for example for nights or weekends, below the first.

​
Temporary coverage

A temporary override hands on-call to one person for a fixed window, for leave, a swap, or an incident. On the schedule, select Add override, choose the Covering person, enter the Reason, the Start time, and the End time, and save. An override wins over every rotation while it lasts. Two overrides of one schedule cannot overlap. The form warns when the person is on call for another schedule at the start, and when the start is already past, which makes the override active at once. End coverage ends a running override early.

​
Schedule details and actions

Open a schedule to see who is on call now and since when, the next handoff, the rotation rules, the temporary coverage, a Next 7 days timeline, and the Change history of the schedule. The menu of a schedule has: Home shows the current shift, and members choose how on-call events reach them in Account settings.

​
Escalation policies

Escalations lists the policies with who is told, and when.
  1. Select New policy and enter the Policy name.
  2. For each step, choose Applies from severity (Any severity, or a minimum severity), the Ack timeout (1, 5, 10, 15, or 30 minutes, or 1 hour), and the Targets: a User, a Schedule, or a Team. Add step adds the next step.
  3. Optionally turn on Repeat the whole policy until acknowledged, with Max loops.
  4. Select Create policy.
A step whose minimum severity is higher than the severity of the incident is skipped. The connectors and the fallback routes of each delivery are set in Settings → Notify → Policies, not in the escalation policy.

​
The default policy

A rule names its escalation policy with escalation_policy_id. A rule that names none, as every rule saved in the rule workbench, uses the alias default: the policy of the organization named default, without regard to case. Without such a policy, its incidents escalate to nobody. Create a policy named default before the first rule, for example with the primary on-call schedule in step 1 and the team lead in step 2. Once an incident has started escalating, it keeps the policy it found, even if the default policy is renamed later.

​
API

Last modified on October 26, 2026