7.5. 灾难恢复警报

记录规则

• record: ramen_sync_duration_seconds

  表达式

  sum by (obj_name, obj_namespace, obj_type, job, policyname)(time() - (ramen_last_sync_timestamp_seconds > 0))

  目的

  卷组最后一次同步时间和时间之间的时间间隔（以秒为单位）。

• Record: ramen_rpo_difference

  表达式

  ramen_sync_duration_seconds{job="ramen-hub-operator-metrics-service"} / on(policyname, job) group_left() (ramen_policy_schedule_interval_seconds{job="ramen-hub-operator-metrics-service"})

  目的

  预期同步延迟和卷复制组执行的实际同步延迟之间的区别。

• record: count_persistentvolumeclaim_total

  表达式

  count(kube_persistentvolumeclaim_info)

  目的

  受管集群中所有 PVC 的总和。

警报

• alert: VolumeSynchronizationDelay

  影响

  critical

  目的

  卷组执行的实际同步延迟是预期的同步延迟。

  YAML

  alert: VolumeSynchronizationDelay
  expr: ramen_rpo_difference >= 3
  for: 5s
  labels:
    severity: critical
  annotations:
    description: "The syncing of volumes is exceeding three times the scheduled snapshot interval, or the volumes have been recently protected. (DRPC: {{ $labels.obj_name }}, Namespace: {{ $labels.obj_namespace }})"
    alert_type: "DisasterRecovery"

• alert: VolumeSynchronizationDelay

  影响

  警告

  目的

  卷组执行的实际同步延迟是预期的同步延迟两倍。

  YAML

  alert: VolumeSynchronizationDelay
  expr: ramen_rpo_difference > 2 and ramen_rpo_difference < 3
  for: 5s
  labels:
    severity: warning
  annotations:
    description: "The syncing of volumes is exceeding two times the scheduled snapshot interval, or the volumes have been recently protected. (DRPC: {{ $labels.obj_name }}, Namespace: {{ $labels.obj_namespace }})"
    alert_type: "DisasterRecovery"

• 警报： WorkloadUnprotected

  影响

  警告

  目的

  应用程序保护状态降级超过 10 分钟

  YAML

  alert: WorkloadUnprotected
  expr: ramen_workload_protection_status == 0
  for: 10m
  labels:
    severity: warning
  annotations:
    description: "Workload is not protected for disaster recovery (DRPC: {{ $labels.obj_name }}, Namespace: {{ $labels.obj_namespace }})."
    alert_type: "DisasterRecovery"