cronjob的限制

如果 startingDeadlineSeconds 未设置（默认），并且 concurrencyPolicy 设置为 Allow，该参数默认为true，则作业可以并行运行。

root@k8s-master01:~# cat test-cronjob.yaml 
apiVersion: batch/v1
kind: CronJob
metadata:
  name: hello
  namespace: test-cronjob
spec:
  schedule: "*/1 * * * *"
  concurrencyPolicy: Allow
  failedJobsHistoryLimit: 2
  successfulJobsHistoryLimit: 2
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: hello
            image: busybox:1.28
            imagePullPolicy: IfNotPresent
            command:
            - /bin/sh
            - -c
            - sleep 100
          restartPolicy: OnFailure

root@k8s-master01:~# kubectl get pod -n test-cronjob
NAME                   READY   STATUS    RESTARTS   AGE
hello-27845595-nmtnc   1/1     Running   0          78s
hello-27845596-xz656   1/1     Running   0          18s

注意：
如果 startingDeadlineSeconds 的设置值低于 10 秒钟，CronJob 可能无法被调度。 这是因为 CronJob 控制器每 10 秒钟执行一次检查。

对于每个 CronJob，CronJob 控制器（Controller） 检查从上一次调度的时间点到现在所错过了调度次数。如果错过的调度次数超过 100 次， 那么它就不会启动这个任务，并记录这个错误:
上述所说为官网所说，本人亲测，1.25版本事件日志会抛出如下警告，但是依然会执行下一个任务。

 Warning  TooManyMissedTimes  4m13s (x6 over 9m13s)   cronjob-controller  (combined from similar events): too many missed start times: 115. Set or decrease .spec.startingDeadlineSeconds or check clock skew

apiVersion: batch/v1
kind: CronJob
metadata:
  name: hello
  namespace: test-cronjob
spec:
  schedule: "*/1 * * * *"
  concurrencyPolicy: Forbid
  failedJobsHistoryLimit: 2
  successfulJobsHistoryLimit: 2
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: hello
            image: busybox:1.28
            imagePullPolicy: IfNotPresent
            command:
            - /bin/sh
            - -c
            - sleep 7000
          restartPolicy: OnFailure

root@k8s-master01:~# kubectl get pod -n test-cronjob
NAME                   READY   STATUS      RESTARTS   AGE
hello-27845626-vk6zj   0/1     Completed   0          120m
hello-27845742-tf8hv   1/1     Running     0          3m42s

root@k8s-master01:~# kubectl describe cronjob hello -n test-cronjob
Name:                          hello
Namespace:                     test-cronjob
Labels:                        <none>
Annotations:                   <none>
Schedule:                      */1 * * * *
Concurrency Policy:            Forbid
Suspend:                       False
Successful Job History Limit:  2
Failed Job History Limit:      2
Starting Deadline Seconds:     <unset>
Selector:                      <unset>
Parallelism:                   <unset>
Completions:                   <unset>
Pod Template:
  Labels:  <none>
  Containers:
   hello:
    Image:      busybox:1.28
    Port:       <none>
    Host Port:  <none>
    Command:
      /bin/sh
      -c
      sleep 7000
    Environment:     <none>
    Mounts:          <none>
  Volumes:           <none>
Last Schedule Time:  Sun, 11 Dec 2022 07:42:00 +0000
Active Jobs:         hello-27845742
Events:
  Type     Reason              Age                     From                Message
  ----     ------              ----                    ----                -------
  Warning  TooManyMissedTimes  13m                     cronjob-controller  too many missed start times: 106. Set or decrease .spec.startingDeadlineSeconds or check clock skew
  Normal   JobAlreadyActive    9m13s (x110 over 118m)  cronjob-controller  Not starting job because prior execution is running and concurrency policy is Forbid
  Warning  TooManyMissedTimes  4m13s (x6 over 9m13s)   cronjob-controller  (combined from similar events): too many missed start times: 115. Set or decrease .spec.startingDeadlineSeconds or check clock skew

如果 startingDeadlineSeconds 字段非空，则控制器会统计从 startingDeadlineSeconds 设置的值到现在而不是从上一个计划时间到现在错过了多少次 Job。 例如，如果 startingDeadlineSeconds 是 300，则控制器会统计在过去 300 秒中错过了多少次 Job。按下述逻辑300s只错过了5次，所以不会抛出警告

apiVersion: batch/v1
kind: CronJob
metadata:
  name: hello1
  namespace: test-cronjob1
spec:
  schedule: "*/1 * * * *"
  concurrencyPolicy: Forbid
  startingDeadlineSeconds: 300
  failedJobsHistoryLimit: 2
  successfulJobsHistoryLimit: 2
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: hello
            image: busybox:1.28
            imagePullPolicy: IfNotPresent
            command:
            - /bin/sh
            - -c
            - sleep 7000
          restartPolicy: OnFailure

root@k8s-master01:~# kubectl get pod -n test-cronjob1
NAME                    READY   STATUS      RESTARTS   AGE
hello1-27845635-52sk9   0/1     Completed   0          117m
hello1-27845751-6cpjk   1/1     Running     0          37s

设置拉参数startingDeadlineSeconds=300，events将不会抛出上述警告

root@k8s-master01:~# kubectl describe cronjob -n test-cronjob1
Name:                          hello1
Namespace:                     test-cronjob1
Labels:                        <none>
Annotations:                   <none>
Schedule:                      */1 * * * *
Concurrency Policy:            Forbid
Suspend:                       False
Successful Job History Limit:  2
Failed Job History Limit:      2
Starting Deadline Seconds:     300s
Selector:                      <unset>
Parallelism:                   <unset>
Completions:                   <unset>
Pod Template:
  Labels:  <none>
  Containers:
   hello:
    Image:      busybox:1.28
    Port:       <none>
    Host Port:  <none>
    Command:
      /bin/sh
      -c
      sleep 7000
    Environment:     <none>
    Mounts:          <none>
  Volumes:           <none>
Last Schedule Time:  Sun, 11 Dec 2022 07:51:00 +0000
Active Jobs:         hello1-27845751
Events:
  Type    Reason            Age                  From                Message
  ----    ------            ----                 ----                -------
  Normal  JobAlreadyActive  2m (x116 over 117m)  cronjob-controller  Not starting job because prior execution is running and concurrency policy is Forbid