I am working on Jenkins Pipeline Script and I have checked-in my jenkinsfile in Git repository and I need to clone to local work space. But by default its cloning to master (Unix) work space but I need it in slave (Windows) work space.
Is there any plugins to change the default Pipeline Script from SCM work space location to slave?
You can do something like this
pipeline {
agent any
options {
skipDefaultCheckout()
}
stages {
stage('checkout') {
steps {
node('windows') {
checkout scm
}
}
}
}
}
OR
pipeline {
agent 'windows'
stages {
stage('build') {
steps {
// build
}
}
}
}
In my case, the following pipeline configuration skips the default checkout on master, and checkout my code just on Jenkins slave.
node {
docker.image('php7.1.30:1.0.0').inside {
skipDefaultCheckout() // this avoid the checkout on master
stage("checkout"){
checkout scm // here the checkout happens on slave node
}
stage('NPM Install'){
sh label: 'NPM INSTALL', script: "npm install"
sh label: 'GRUNT INSTALL', script: "npm install -g grunt-cli"
}
stage('Executing grunt') {
sh label: 'GRUNT DEFAULT', script: "grunt default"
}
}
}
Related
I have created very basic Multibranch Pipeline on my local Jenkins via BlueOcean UI. From default config I removed almost all behaviors except one for discovering branches. The config looks line follows:
Within Jenkinsfile I'm trying to setup following scenario:
Checkout branch
(optionally) Merge it to master branch
Build Back-end
Build Front-end
Snippet from my Jenkinsfile:
pipeline {
agent none
stages {
stage('Setup') {
agent {
label "master"
}
steps {
sh "git checkout -f ${env.BRANCH_NAME}"
}
}
stage('Merge with master') {
when {
not {
branch 'master'
}
}
agent {
label "master"
}
steps {
sh 'git checkout -f origin/master'
sh "git merge --ff-only ${env.BRANCH_NAME}"
}
}
stage('Build Back-end') {
agent {
docker {
image 'openjdk:8'
}
}
steps {
sh './gradlew build'
}
}
stage ('Build Front-end') {
agent {
docker {
image 'saddeveloper/node-chromium'
}
}
steps {
dir ('./front-end') {
sh 'npm install'
sh 'npm run buildProd'
sh 'npm run testHeadless'
}
}
}
}
}
Pipeline itself and building steps works fine, but the problem is that Jenkins adds "Check out from version control" step before each stage. The step looks for new branches, fetches refs, but also checks out current branch. Here is relevant output from full build log:
// stage Setup
> git checkout -f f067047bbdd3a5d5f9d1f2efae274bc175829595
sh git checkout -f my-branch
// stage Merge with master
> git checkout -f f067047bbdd3a5d5f9d1f2efae274bc175829595
sh git checkout -f origin/master
sh git merge --ff-only my-branch
// stage Build Back-end
> git checkout -f f067047bbdd3a5d5f9d1f2efae274bc175829595
sh ./gradlew build
// stage Build Front-end
> git checkout -f f067047bbdd3a5d5f9d1f2efae274bc175829595
sh npm install
sh npm run buildProd
sh npm run testHeadless
So as you see it effectively resets working directory to particular commit before every stage git checkout -f f067...595.
Is there any way to disable this default checkout behavior?
Or any viable option how to implement such optional merging to master branch?
Thanks!
By default, git scm will be executed in a Jenkins pipeline. You can disable it by doing:
pipeline {
agent none
options {
skipDefaultCheckout true
}
...
Also, I'd recommend take a look to other useful pipeline options https://jenkins.io/doc/book/pipeline/syntax/#options
I have the following advanced scripted pipeline in a Jenkinsfile:
stage('Generate') {
node {
checkout scm
}
parallel windows: {
node('windows') {
sh 'cmake . -Bbuild.windows -A x64'
}
},
macos: {
node('apple') {
sh '/usr/local/bin/cmake . -DPLATFORM="macos" -Bbuild.macos -GXcode'
}
},
ios: {
node('apple') {
sh '/usr/local/bin/cmake . -DPLATFORM="ios" -Bbuild.ios -GXcode'
}
}
}
Note the top node that precedes the parallel windows/macos/ios nodes. Does this mean that checkout scm will be invoked on every subsequent building node (windows/apple), before proceeding to the parallel steps? In other words, does the script above guarantee that the repository will be checked out on every node that will be involved at any stage of this build?
Many thanks.
The first node step will allocate any build agent and check out the source code.
Later, additional nodes will be allocated, where I can promise you that cmake will fail, as it works with an empty directory.
You can use stash and unstash to copy over the files that are needed for the build (and subsequent stages):
stage('Generate') {
node {
checkout scm
stash 'source'
}
parallel windows: {
node('windows') {
unstash 'source'
sh 'cmake . -Bbuild.windows -A x64'
}
},
macos: {
node('apple') {
unstash 'source'
sh '/usr/local/bin/cmake . -DPLATFORM="macos" -Bbuild.macos -GXcode'
}
},
ios: {
node('apple') {
unstash 'source'
sh '/usr/local/bin/cmake . -DPLATFORM="ios" -Bbuild.ios -GXcode'
}
}
}
My project has a Jenkinsfile that runs smoothly. The problem is that I need to run some commands only on certain occasions. I'm using the Github plugin. I need to run the deploy only when it is in the master or a new tag, one will be for staging and the other will be production.
pipeline {
agent any
stages {
stage('Test') {
steps {
sh 'node -v'
sh 'yarn install'
sh 'yarn test -- --coverage'
}
}
stage('Build') {
steps {
sh 'yarn build'
}
}
stage('Deploy') {
steps {
sh 'aws s3 sync ./build s3://my.bucket --only-show-errors'
}
}
}
}
I need the master to deploy to a bucket and when it is new tag to another. How can I create this conditional?
How about the following working as two conditionals for two separate deployment scenarios? I think it's better to work with this using variables to indicate deployment scenarios instead of splitting this to two distinctly different steps though. You could for example write a shell script that would handle everything inside depending on tags/branches/whatever you need instead of forcing yourself to control this on pipeline level.
Each stage will have it's steps executed only when when part is satisfied. Stage Deploy will only work for master branch, while stage Deploy_NonMaster will only work any non master branch. Using the method written in when conditionals you can check for anything, including tags or whatnot.
stage ('Deploy') {
when {
expression {
GIT_BRANCH = sh(returnStdout: true, script: 'git rev-parse --abbrev-ref HEAD').trim()
return (GIT_BRANCH == 'master')
}
}
steps {
echo 'Do stuff/deploy.'
}
}
stage ('Deploy_NonMaster') {
when {
expression {
GIT_BRANCH = sh(returnStdout: true, script: 'git rev-parse --abbrev-ref HEAD').trim()
return !(GIT_BRANCH == 'master')
}
}
steps {
echo 'Do stuff/deploy.'
}
}
I'm trying to create some Docker images. For that I want to use the version number specified in the Maven pom.xml file as tag. I am however rather new to the declarative Jenkins pipelines and I can't figure out how to change my environment variable so that VERSION contains the right version for all stages.
This is my code
#!groovy
pipeline {
tools {
maven 'maven 3.3.9'
jdk 'Java 1.8'
}
environment {
VERSION = '0.0.0'
}
agent any
stages {
stage('Checkout') {
steps {
git branch: 'master', credentialsId: '290dd8ee-2381-4c5b-8d33-5631d03ee7be', url: 'git#gitlab.crosslang.local:company/SOME-API.git'
sh "git clean -f && git reset --hard origin/master"
}
}
stage('Build and Test Java code') {
steps {
script {
def pom = readMavenPom file: 'pom.xml'
VERSION = pom.version
}
echo "${VERSION}"
sh "mvn clean install -DskipTests"
}
}
stage('Build Docker images') {
steps {
dir('whales-microservice/src/main/docker'){
sh 'cp ../../../target/whales-microservice-${VERSION}.jar whales-microservice.jar'
script {
docker.build "company/whales-microservice:${VERSION}"
}
}
}
}
}
}
The problem is the single quote of the statement
sh 'cp ../../../target/whales-microservice-${VERSION}.jar whales-microservice.jar'
single quotes don't expand variables in groovy: http://docs.groovy-lang.org/latest/html/documentation/#_string_interpolation
so you have to double quote your shell statement:
sh "cp ../../../target/whales-microservice-${VERSION}.jar whales-microservice.jar"
I just wanted to mention that if you have pipeline-utility-steps plugin installed you can use readMavenPom() in the environment part, too. It looks like this:
environment {
VERSION = readMavenPom().getVersion()
}
When building a multibranch pipeline, I send each of my projects to SonarQube using the SonarQube plugin like so:
pipeline {
agent any
options {
buildDiscarder(logRotator(numToKeepStr:'20'))
timeout(time: 30, unit: 'MINUTES')
}
tools {
maven 'Maven 3.3.9'
jdk 'JDK 1.8'
}
stages {
stage('Checkout') {
steps {
echo 'Checking out..'
checkout scm
echo "My branch is: ${env.BRANCH_NAME}"
}
}
stage('Build') {
steps {
echo 'Building..'
bat 'mvn clean verify -P!local'
}
}
stage('SonarQube analysis'){
steps{
echo 'Analysing...'
withSonarQubeEnv('SonarQube') {
bat 'mvn org.sonarsource.scanner.maven:sonar-maven-plugin:3.2:sonar'
}
}
}
}
}
It works fine, but one thing I need it to do is to change the name of the project in SonarQube to be projectName/builtBranch instead of just the project name. Is there a way I can do this using the pipeline?
This doesn't seem to be a Jenkins issue; rather you should be able to set the various sonar.* properties (e.g. sonar.projectName or sonar.branch) when running the Maven plugin.
The documentation seems to have a full list:
https://docs.sonarqube.org/display/SONAR/Analysis+Parameters