Alertmanager中可以使用repeat_interval选项指定在一个告警重复发送前必须等待的时间间隔。可以通过添加一条额外的规则来实现基于时间的升级机制，该规则定义了在指定的时间后发送进一步的警报通知，直到警报被解决。以下是示例代码：

groups:

• name: example rules:

  Base alert rule

  • alert: ServiceFailed expr: job_failed{job="example-service"} > 0 for: 5m labels: severity: page

  Time-based escalation

  • alert: ServiceFailedEscalation expr: job_failed{job="example-service"} > 0 for: 10m annotations: message: "Service failed for more than 10 minutes!" labels: severity: page

    Send a new notification every 2 minutes until resolved

    repeat_interval: 2m

    Escalate to a higher severity after 20 minutes

    and then every 10 minutes thereafter

    routes:
    • match: severity: page repeat_interval: 10m routes:
      • match: severity: critical continue: true

在上面的示例中，“ServiceFailed”是基本告警规则。如果此规则持续不断地（每5分钟）触发，则将启动“ServiceFailedEscalation”规则。这条规则包含一个repeate_interval选项，该选项每2分钟发送一次通知，直到警报被解决。如果此问题在10分钟内未解决，则“ServiceFailedEscalation”将通过重复检查“job_failed”表达式来检查旧警报，并以10分钟为时间间隔将警报转发到“critical”紧急程度。