前言
容器分为两类
- 在线业务 - 服务类容器
须要继续不中断的提供服务，容器须要始终运行
- 离线业务 - 工作类容器
个别执行一次性工作，比方统计日志数据等，运行实现后容器即可敞开
更新历史
- 20200618 - 初稿 - 左程立
- 原文地址 - https://blog.zuolinux.com/2020/06/18/controller-job-cronjob.html
Job
cat job.yaml
apiVersion: batch/v1kind: Jobmetadata: name: job-testspec: template: metadata: name: job-test spec: containers: - name: test-job image: busybox command: ["echo", "test job!"] restartPolicy: Never参数
restartPolicy 只能是 Never 或者 onFailure
启动
[root@master01 ~]# kubectl apply -f job.yaml job.batch/job-test created运行状态
[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-t2gbw 0/1 ContainerCreating 0 12s[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-t2gbw 0/1 Completed 0 24s查看运行后果
[root@master01 ~]# kubectl logs job-test-t2gbwtest job!Job 执行失败如何解决
将 command 胡乱设置导致 job 无奈胜利启动
command: ["echo123", "test job!"] restartPolicy: Never 当 restartPolicy: Never 时，能够看到 k8s 在一直的开新的 pod
但不会让它始终开上来，默认参数 spec.backoffLimit: 6 会进行阻止
[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-2zrt4 0/1 ContainerCannotRun 0 87sjob-test-5z884 0/1 ContainerCannotRun 0 77sjob-test-cxgmm 0/1 ContainerCannotRun 0 107sjob-test-nt9gt 0/1 ContainerCannotRun 0 57sjob-test-wxv7l 0/1 ContainerCreating 0 17s当 restartPolicy: OnFailure 时，k8s 不开新 pod，只会一直重启 pod
[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-bd695 0/1 CrashLoopBackOff 4 3m15s如果 job 没有报错，然而它始终不完结怎么办？
参数 spec.activeDeadlineSeconds: 100 会在 100s 后将 Job 中所有 Pod 进行敞开此时 Pod 敞开状态为 reason: DeadlineExceeded并行 Job
cat job.yaml
apiVersion: batch/v1kind: Jobmetadata: name: job-testspec: parallelism: 2 ......参数
parallelism: 2 仅容许最多两个 Pod 同时运行
查看运行状况
[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-6ttz9 0/1 ContainerCreating 0 8sjob-test-mknjc 0/1 ContainerCreating 0 8s[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-6ttz9 0/1 Completed 0 68sjob-test-mknjc 0/1 Completed 0 68s[root@master01 ~]# kubectl get jobNAME COMPLETIONS DURATION AGEjob-test 0/1 of 2 10s 10s[root@master01 ~]# kubectl get jobNAME COMPLETIONS DURATION AGEjob-test 2/1 of 2 19s 66s减少参数 completions
apiVersion: batch/v1kind: Jobmetadata: name: job-testspec: parallelism: 2 completions: 8参数
completions: 8 示意至多胜利运行 8 个 pod
如果不是8个胜利，那么会依据 restartPolicy 的策略进行解决
能够认为是一种查看机制
查看运行状况
[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-6xwpt 0/1 ContainerCreating 0 8sjob-test-grk4q 0/1 ContainerCreating 0 8s[root@master01 ~]# kubectl get jobNAME COMPLETIONS DURATION AGEjob-test 0/8 11s 11s[root@master01 ~]# kubectl get jobNAME COMPLETIONS DURATION AGEjob-test 6/8 76s 76s[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEjob-test-2jxqx 0/1 Completed 0 28sjob-test-6xwpt 0/1 Completed 0 90sjob-test-gndq7 0/1 Completed 0 70sjob-test-grk4q 0/1 Completed 0 90sjob-test-n96k6 0/1 Completed 0 70sjob-test-np4n7 0/1 Completed 0 49sjob-test-scc9c 0/1 Completed 0 28sjob-test-zcnbp 0/1 Completed 0 49s[root@master01 ~]# kubectl get jobNAME COMPLETIONS DURATION AGEjob-test 8/8 82s 91sCronJob
CronJob 会定时 create 一个 job 对象
cat cronjob.yaml
apiVersion: batch/v1beta1kind: CronJobmetadata: name: hellospec: schedule: "*/1 * * * *" jobTemplate: spec: template: spec: containers: - name: hello image: busybox command: ["echo","test cron job!"] restartPolicy: OnFailure查看运行后果
[root@master01 ~]# kubectl get cronjobNAME SCHEDULE SUSPEND ACTIVE LAST SCHEDULE AGEhello */1 * * * * False 0 67s 3m45s[root@master01 ~]# kubectl get jobsNAME COMPLETIONS DURATION AGEhello-1595319480 1/1 27s 3m23shello-1595319540 1/1 19s 2m23shello-1595319600 1/1 24s 83shello-1595319660 0/1 22s 22s[root@master01 ~]# kubectl get podNAME READY STATUS RESTARTS AGEhello-1595319480-2kl8h 0/1 Completed 0 3m21shello-1595319540-gwthv 0/1 Completed 0 2m21shello-1595319600-54548 0/1 Completed 0 81shello-1595319660-8xqn6 0/1 ContainerCreating 0 20s能够看到，就是把后面说过的 Job 对象进行定时的运行。
结束语
Job/CronJob 控制器能够进行离线业务的解决，不须要长期运行一个 Pod。
这类控制器治理的 Pod 不对外提供继续在线服务，工作运行实现后，即会将容器进行敞开，能够用于一些定时数据处理类的工作。
分割我
微信公众号：zuolinux_com