在 Jenkins 管道中运行并行 kubernetes 作业

Question

I'm running perfomance tests on Jenkins. Test that may include multiple instances of the same container to generate necessary load. I can't hardcode number of instances as it varies based on params for tests. I've tried to use the following code:

pipeline {
agent any
stages {
  stage('Running Jmeter') {
    agent {
      kubernetes {
          label "jmeter_tests_executor"
          yaml '''
          apiVersion: batch/v1
          kind: Job
          metadata:
            name: jmeter
            namespace: jenkins
          spec:
            parallelism: 2 
            backoffLimit: 1 
            ttlSecondsAfterFinished: 100
          ...

But it doesn't work. It's hanging on pod scheduling(jobs works ok if you apply this manifest directly on kubernetes cluster without Jenkins).
If someone had experience with it, please share your workarounds or ideas how to implement this idea.

Answer 1

Maybe try something like this

stage("RUN LOAD TEST") {
    steps {
        script {
            //params.each creates an array of stages
            paramsToTest.each {param -> 
                load["load test"] = {
                    stage("Executing run ${param}") {
                        agent {
                            kubernetes {
                                label "jmeter_tests_executor"
                                yaml '''
                                apiVersion: batch/v1
                                kind: Job
                                metadata:
                                name: jmeter
                                namespace: jenkins
                                spec:
                                parallelism: 2 
                                backoffLimit: 1 
                                ttlSecondsAfterFinished: 100
                                ...
                                '''
                            }    
                        }
                        steps {
                            <EXECUTE LOAD TEST COMMAND>
                        }
                    }
                }
                parallel(load) //actually executes the parallel stages
            }
        }
    }
}

What this does is use an array of something and then generates stages based on that array. The agent params in the stage should tell Jenkins to create a new pod with each execution in parallel.