Kubernetes k8s作业已完成，但pod显示；“OOMKILLED”；

我运行的作业内存限制为2Gi，这似乎不够。作业的状态显示为“已完成”，1成功

status:
  completionTime: "2020-05-09T03:44:07Z"
  conditions:
  - lastProbeTime: "2020-05-09T03:44:07Z"
    lastTransitionTime: "2020-05-09T03:44:07Z"
    status: "True"
    type: Complete
  startTime: "2020-05-09T03:42:07Z"
  succeeded: 1

而pod的状态为“OOMKILLED”

---

提高内存限制可以解决这个问题，但我不知道这是怎么发生的。这让我觉得作业状态有时可能是错误的？

pod阶段显示“成功”，这是怎么发生的？它不是因为OOMKILL而被杀死一次，然后在另一个有足够空间的节点中重新执行，直到成功吗？

status:
  conditions:
  - lastProbeTime: null
    lastTransitionTime: "2020-05-09T03:42:07Z"
    reason: PodCompleted
    status: "True"
    type: Initialized
  - lastProbeTime: null
    lastTransitionTime: "2020-05-09T03:44:07Z"
    reason: PodCompleted
    status: "False"
    type: Ready
  - lastProbeTime: null
    lastTransitionTime: "2020-05-09T03:44:07Z"
    reason: PodCompleted
    status: "False"
    type: ContainersReady
  - lastProbeTime: null
    lastTransitionTime: "2020-05-09T03:42:07Z"
    status: "True"
    type: PodScheduled
  containerStatuses:
  - containerID: docker://50db639d6f29a56f12b0b878a2a5ee957a9e36fb6dcd089831e4422b85a42e3a
    lastState: {}
    name: fairing-job
    ready: false
    restartCount: 0
    state:
      terminated:
        containerID: docker://50db639d6f29a56f12b0b878a2a5ee957a9e36fb6dcd089831e4422b85a42e3a
        exitCode: 0
        finishedAt: "2020-05-09T03:44:07Z"
        reason: OOMKilled
        startedAt: "2020-05-09T03:42:13Z"
  hostIP: 10.201.1.202
  phase: Succeeded
  podIP: 10.178.140.83
  qosClass: Guaranteed
  startTime: '2020-05-09T07:38:01Z'