CODEWITHSUNDEEP
jenkinsgroovyjenkins-pipelinejenkins-job-dsl
Mark McLaren
Mark McLaren

Reputation: 11540

Passing maps into Jenkins pipeline jobs

I have a Jenkins Job DSL seed job that calls out to a couple of pipeline jobs e.g.

pipelineJob("job1") {
    definition {
        cps {
            script(readFileFromWorkspace('job1.groovy'))
        }
        parameters {
            choiceParam('ENV', ['dev', 'prod'], 'Build Environment')
         }
    }
}
pipelineJob("job2") {
    definition {
        cps {
            script(readFileFromWorkspace('job2.groovy'))
        }
        parameters {
            choiceParam('ENV', ['dev', 'prod'], 'Build Environment')
         }
    }
}

job1.groovy and job2.groovy are standard Jenkinsfile style pipelines.

I want to pass a couple of common maps into these jobs. These contains things that may vary between environments, e.g. target servers, credential names.

Something like:

def SERVERS_MAP = [
'prod': [
        'prod-server1',
        'prod-server2',
      ],
'dev': [
        'dev-server1',
        'dev-server2',
      ],
]

Can I define a map in my seed job that I can then pass and access as a map in my pipeline jobs?

Upvotes: 4

Views: 9921

Answers (2)

yu jiang
yu jiang

Reputation: 79

You can use string parameter pass the Map val, downstream read it as json format.

UPSTREAM PIPELINE

timestamps{
node("sse_lab_CI_076"){ //${execNode}
    currentBuild.description="${env.NODE_NAME};"

    stage("-- regression execute --"){
        def test_map = 
        """
            {
                "gerrit_patchset_commit": "aad5fce",
                "build_cpu_x86_ubuntu":     [
                    "centos_compatible_build_test",
                    "gdb_compatible_build_test",
                    "visual_profiler_compatible_build_test"
                ],
            }
        """
        build(job: 'tops_regression_down',
            parameters: [string(name: 'UPSTREAM_JOB_NAME',
                                value: "${env.JOB_BASE_NAME}"),
                        string(name: 'UPSTREAM_BUILD_NUM',
                                value: "${env.BUILD_NUMBER}"),
                        string(name: 'MAP_PARAM',
                                value: "${test_map}"),
                        ],
            propagate: true,
            wait: true)
    }
}

}

DOWNSTREAM PIPELINE

timestamps{
node("sse_lab_inspur_076"){ //${execNode}
    currentBuild.description="${env.NODE_NAME};"
    stage('--in precondition--'){
        dir('./'){
            cleanWs()
            println("hello world")
            println("${env.MAP_PARAM}")
            Map result_json = readJSON(text: "${env.MAP_PARAM}")
            println(result_json)
        }
    }
}

}

Upvotes: 0

Mark McLaren
Mark McLaren

Reputation: 11540

I've come up with a hacky workaround using the pipeline-utility-steps plugin.

Essentially I pass my data maps around as JSON.

So my seed job might contain:

def SERVERS_MAP = ''' 
{
  "prod": [
    "prod-server1",
    "prod-server2"
  ],
  "dev": [
    "dev-server1",
    "dev-server2"
  ]
}
'''

pipelineJob("job1") {
    definition {
        cps {
            script(readFileFromWorkspace('job1.groovy'))
        }
        parameters {
            choiceParam('ENV', ['dev', 'prod'], 'Build Environment')
            stringParam('SERVERS_MAP', "${SERVERS_MAP}", "")
         }
    }
}

and my pipeline would contain something like:

def serversMap = readJSON text: SERVERS_MAP
def targetServers = serversMap["${ENV}"]
targetServers.each { server ->
   echo server
}

I could also extract these variables into a JSON file and read them from there.

Although it works, it feels wrong somehow.

Upvotes: 5

Related Questions