26

Jenkins 2 has pipelines has a first class citizen. However, in the examples the tasks seem to be executed as a single sequence:

node {
   // Mark the code checkout 'stage'....
   stage 'Checkout'

   // Get some code from a GitHub repository
   git url: 'git@github.com:elifesciences/elife-bot.git'

   // Mark the code build 'stage'....
   stage 'Build'
   echo "Unit tests will run here"

   stage "Production"
   echo "Deploying to production environment"
}

For deployment into production system it's often useful to require manual approval; is there a way to insert a manual button to press inside a pipeline?

I have been looking for possible steps to accomplish this on the docs, to no avail.

2
  • I don't know Jenkins, but isn't there a way to split your build plan into several steps, and having some of these steps be run only on a "manual trigger"?
    – tiktak
    May 5, 2016 at 16:06
  • Best partial solution so far: an input step in the pipeline which stops and asks the user for input (or to abort the build). However, the stage and the status indicator keeps flashing while I wanted a stable state (e.g. you get into it Friday afternoon and decide to deploy on Monday.) May 6, 2016 at 12:19

5 Answers 5

23

input is the option you are looking for. Here is the way I'm using it. It's important to have the step outside a node, otherwise jenkins will hold an agent waiting for the next step. Keep in mind the second node may not use the same workspace as the first.

node {
    stage('build'){
        echo "building"
    }
}
stage('Deploy approval'){
    input "Deploy to prod?"
}
node {
    stage('deploy to prod'){
        echo "deploying"
    }
}
7
  • Given multiple pipelines can get there, what happens to the older ones who do not get deployed to production? Is there a way to stop the older ones from remaining there (don't know if they will be flashing) in an incomplete state? Feb 17, 2017 at 15:57
  • 1
    as far as i can tell it will flash forever until you click abort, which is pretty crappy. you could probably setup a timeout to prevent some of these from getting lost. After the timeout you would lose the ability to deploy it though. jenkins.io/doc/pipeline/steps/workflow-basic-steps/… Feb 17, 2017 at 17:56
  • 2
    I didn't understand that input could be configured to not hold up an agent. This makes input way more useful. Dec 6, 2017 at 18:30
  • 1
    Would be nice to have the possibility to redeploy a version, without building, or to deploy the previous version.
    – tehnicaorg
    Jul 18, 2019 at 12:50
  • It would be nice if one could specify who is allowed to give the approval, perhaps based the internal jenkins roles. Oct 27, 2020 at 20:46
3

I did in as shown below way by reading this docs https://jenkins.io/doc/book/pipeline/syntax/

pipeline {
environment {
    BRANCH_NAME = "${env.BRANCH_NAME}"
}
agent any
stages{
    stage('Build-Initiator-Info'){
            steps{
                sh 'echo "Send Info"'
            }
    }
    stage('Build') {
        steps{
             catchError {
                sh 'echo "This is build"'
            }
         }
         post {
            success {
                echo 'Compile Stage Successful . . .'
            }
            failure {
                echo 'Compile stage failed'
                error('Stopping early…')

             }
    }
   }
  stage ('Deploy To Prod'){
  input{
    message "Do you want to proceed for production deployment?"
  }
    steps {
                sh 'echo "Deploy into Prod"'

              }
        }
  }
   }
2

Additionally, You can also add auto-timeout like below

        stage('build') {
        steps {
            sh  """
                # Some commands
                """
            script {
              timeout(time: 10, unit: 'MINUTES') {
                input(id: "Deploy Gate", message: "Deploy ${params.project_name}?", ok: 'Deploy')
              }
            }
        }
    }

    stage('deploy') {
        when {
            branch 'master'
        }
        steps {
            sh  """
                # some commands
                """
        }
    }

If you look it up you can also bind the jenkins input to credentials of users accessing Jenkins if you only wish to permit specific individuals to be capable of answering - it also is underpinned by the fact that your Git controls are sufficient too.

1
  • Thank you. I tried a couple of syntaxes for input, but this was the only one that ended up working for me.
    – Sam
    Oct 13, 2021 at 21:15
1

In the end I created separate test-project and prod-project pipelines, where at the end of test-project the code is merged into an approved branch.

Then, the prod-project pipeline can be set up not to trigger for every new commit so that it can be deployed on-demand.

2
  • 1
    This does not answer the question at all Oct 18, 2020 at 6:28
  • 2
    This is my own question, so I shared the workaround I did to solve the problem. Oct 26, 2020 at 17:30
0

This is just a simple example but you can trigger it however you need.

stage{
    script{
        input "Continue?"
        ...enter code here
        ...
    }
}

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.