Create N jobs with Kubernetes-CodePudding

I have the following Kubernetes job defined in job.yaml

apiVersion: batch/v1
kind: Job
metadata:
  name: pi-$count
spec:
  template:
    spec:
      containers:
      - name: pi
        image: perl:5.34.0
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never
  backoffLimit: 4

I apply this with the following bash script:

for i in {1..3}
do
    kubectl apply -f job.yaml
done

Is there a way to create 3 jobs here and use an environment variable pi-$count so pi-1, pi-2, pi-3 are created?

CodePudding user response：

You can use sed to replace the $count and create a manifest file for each job and apply it.

For example create a file called pi-job.yaml (I'm taking your code as example)

apiVersion: batch/v1
kind: Job
metadata:
  name: pi-$count
spec:
  template:
    spec:
      containers:
      - name: pi
        image: perl:5.34.0
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never
  backoffLimit: 4

Create a folder called job to store the manifest files (this directory is completely optional)You can use something like this

    mkdir job
    for i in {1..3}
    do
       cat pi-job.yaml | sed "s/\$count/$i/" > ./jobs/job-$i.yaml;
       kubectl apply -f ./jobs/job-$i.yaml
    done

By executing this code 3 jobs will be created in the name of pi-1,pi-2 and pi-3. But the manifest files will reside in the job folder. You need to clean it up. Otherwise you can add one more line to the script to remove the files like this

 mkdir job
 for i in {1..3}
 do
   cat pi-job.yaml | sed "s/\$count/$i/" > ./jobs/job-$i.yaml;
   kubectl apply -f ./jobs/job-$i.yaml
   rm -rf ./jobs/job-$i.yaml
 done

For more detailed information refer to the official k8's job document

CodePudding user response：

Use sed command to replace the job counter (COUNT) and create temporary files for each of the job instances. You can delete all the temporary job yamls after deployment. In this way you can keep your original job yaml file intact for reusing. Let's say I am using your kubernetes job yaml named 'job.yaml'.

apiVersion: batch/v1
kind: Job
metadata:
  name: pi-COUNT
spec:
  template:
    spec:
      containers:
      - name: pi
        image: perl:5.34.0
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never
  backoffLimit: 4

Modify your bash script like the following. I'm naming it jobrunner.sh.

#!/usr/bin/bash
for i in {1..3}
do
    sed 's|COUNT|'${i}'|g' job.yaml > job-$i.yaml
    kubectl apply -f job-$i.yaml
done
rm job-*.yaml

Run this bash script and 3 jobs will be created in your Kubernetes named pi-1, pi-2 & pi-3. You can comment out the last part where the temporary job files are being deleted with rm job-*.yaml command if you wish to keep those files.