“Learning k8s from Scratch”“
Introduction to Job
Job is primarily responsible forbatch processing (processing a specified number of tasks at once)temporaryone-time (each task runs only once and then ends)tasks. The characteristics of Job are as follows:
* When the pod created by Job successfully finishes, Job will record the number of successfully finished pods.
* When the number of successfully finished pods reaches the specified amount, Job will complete execution.
Resource manifest file for Job:
apiVersion:batch/v1 # Version kind:Job # Type metadata: # Metadata name: # Job name namespace: # Namespace labels: # Labels controller:job spec: # Description details completions:1 # Specifies the number of successful pod runs required for the job, default value: 1 parallelism:1 # Specifies the number of pods to run concurrently, default value: 1 activeDeadlineSeconds:30 # Specifies the time limit for the job to run, if it exceeds this time and has not finished, the system will attempt to terminate it backoffLimit:6 # Specifies the number of retries after job failure, default: 6 manualSelector:true # Whether to use selector to choose pods, default: false selector: # Selector, through which to specify which pods this controller manages matchLabels: # Labels matching rules app:counterpod matchExpressions: # Expressions matching rules -{key:app,operator:In,values:[counterpod]} template: # Template, when the number of replicas is insufficient, it will create pod replicas based on the template below metadata: labels: app:counterpod spec: restartPolicy:Never# Restart policy can only be set to Never or OnFailure containers: -name:counter image:busybox:1.34.1 command:["bin/sh","-c","for i in 9 8 7 6 5 4 3 2 1;do echo $i; sleep 2; done"]Explanation of restart policy settings::
* If set toOnFailure, then the job will restart the container when the pod fails, rather than creating a new pod, and the failed count remains unchanged.
* If set toNever, then the job will create a new pod when the existing pod fails, and the failed pod will not disappear, nor will it restart, and the failed count increases by 1.
* If set toAlways, it means to always restart, which means the job task will be executed repeatedly, of course, this is incorrect, so it cannot be set toAlways.
Creating pc-job.yaml:
cat > pc-job.yaml <<EOF apiVersion:batch/v1 kind:Job metadata: name:pc-job namespace:dev spec: manualSelector:true selector: matchLabels: app:counter-pod template: metadata: labels: app:counter-pod spec: restartPolicy:Never containers: - name:counter image:busybox:1.34.1 command:["bin/sh","-c","for i in 9 8 7 6 5 4 3 2 1; do echo $i; sleep 3; done"] EOF# Open 2 shells#1. Observe job statuskubectl get job -n dev -o wide -wNAME COMPLETIONS DURATION AGE CONTAINERS IMAGES SELECTORpc-job 0/1 7s 7s counter busybox:1.34.1 app=counter-podpc-job 1/1 37s 37s counter busybox:1.34.1 app=counter-pod# Create jobkubectl create -f pc-job.yaml# Delete jobkubectl delete -f pc-job.yaml# Modify pcjob.yaml, adjust the total number of pods and parallelism by setting the following two options under spec:#completions:6# Specifies the job needs to successfully run Pods a total of 6 times #parallelism:3# Specifies the job to run Pods concurrently as 3cat > pc-job.yaml << EOF apiVersion:batch/v1 kind:Job metadata: name:pc-job namespace:dev spec: manualSelector:true completions:6 parallelism:3 selector: matchLabels: app:counter-pod template: metadata: labels: app:counter-pod spec: restartPolicy:Never containers: - name:counter image:busybox:1.34.1 command:["bin/sh","-c","for i in 9 8 7 6 5 4 3 2 1; do echo $i; sleep 3; done"] EOF# Open 2 shells#1. Observe job statuskubectl get job -n dev -o wide -wNAME COMPLETIONS DURATION AGE CONTAINERS IMAGES SELECTORpc-job 0/6 0s 0s counter busybox:1.34.1 app=counter-podpc-job 0/6 0s 0s counter busybox:1.34.1 app=counter-podpc-job 1/6 29s 29s counter busybox:1.34.1 app=counter-podpc-job 2/6 29s 29s counter busybox:1.34.1 app=counter-podpc-job 3/6 29s 29s counter busybox:1.34.1 app=counter-podpc-job 4/6 58s 58s counter busybox:1.34.1 app=counter-podpc-job 5/6 58s 58s counter busybox:1.34.1 app=counter-podpc-job 6/6 58s 58s counter busybox:1.34.1 app=counter-pod#2. Observe pod status, at this point, you will find that the job will run 3 pods each time, executing a total of 6 pods kubectl get pods -n dev -wNAME READY STATUS RESTARTS AGEpc-job-nc82v 0/1 Pending 0 0spc-job-nc82v 0/1 Pending 0 0spc-job-v7bft 0/1 Pending 0 0spc-job-9kwrt 0/1 Pending 0 0spc-job-v7bft 0/1 Pending 0 0spc-job-9kwrt 0/1 Pending 0 0spc-job-nc82v 0/1 ContainerCreating 0 0spc-job-v7bft 0/1 ContainerCreating 0 0spc-job-9kwrt 0/1 ContainerCreating 0 0spc-job-v7bft 0/1 ContainerCreating 0 1spc-job-nc82v 0/1 ContainerCreating 0 1spc-job-9kwrt 0/1 ContainerCreating 0 1spc-job-v7bft 1/1 Running 0 1spc-job-9kwrt 1/1 Running 0 2spc-job-nc82v 1/1 Running 0 2spc-job-v7bft 0/1 Completed 0 29spc-job-x47lm 0/1 Pending 0 0spc-job-x47lm 0/1 Pending 0 0spc-job-x47lm 0/1 ContainerCreating 0 0spc-job-9kwrt 0/1 Completed 0 29spc-job-s8w5m 0/1 Pending 0 0spc-job-s8w5m 0/1 Pending 0 0spc-job-nc82v 0/1 Completed 0 29spc-job-hxnj5 0/1 Pending 0 0spc-job-hxnj5 0/1 Pending 0 0spc-job-s8w5m 0/1 ContainerCreating 0 0spc-job-hxnj5 0/1 ContainerCreating 0 0spc-job-x47lm 0/1 ContainerCreating 0 0spc-job-s8w5m 0/1 ContainerCreating 0 1spc-job-hxnj5 0/1 ContainerCreating 0 1spc-job-x47lm 1/1 Running 0 2spc-job-hxnj5 1/1 Running 0 2spc-job-s8w5m 1/1 Running 0 2spc-job-x47lm 0/1 Completed 0 29spc-job-hxnj5 0/1 Completed 0 29spc-job-s8w5m 0/1 Completed 0 29s# Create jobkubectl create -f pc-job.yaml# Delete jobkubectl delete -f pc-job.yaml