-
Type:
New Feature
-
Resolution: Unresolved
-
Priority:
Major
-
Component/s: pipeline-model-definition-plugin
If you have a couple of microservices inside a single git repository and for each microservice there is a Jenkins Job sharing the same Jenkinsfile. e. g. https://github.com/Xaseron/phippyandfriends
The goal is that the pipeline for each microservice is only run if a change is made in the subdirectory or the build is triggered manually.
The current possible solution is kind of ugly: (the job is multibranch pipeline named after the microservice e.g. captainkube)
pipeline {
options {
timestamps()
}
agent {
node {
label 'master'
}
}
environment {
/* get the foo out of e.g. foo/master */
MICROSERVICE = "${env.JOB_NAME.tokenize('/')[0]}"
}
stages {
/* useless stage to get the when block at the top level */
stage('Meta') {
/* only start a build when something changed inside the microservice or it is triggered manually */
when {
anyOf {
triggeredBy cause: 'UserIdCause'
changeset "$MICROSERVICE/**"
}
}
stages {
stage('Build') {
steps {
echo "I've got a change in $MICROSERVICE"
echo "docker build -t $MICROSERVICE:$BUILD_NUMBER $MICROSERVICE"
}
}
stage('Test') {
steps {
echo 'test'
}
}
stage('Deploy') {
steps {
echo 'deploy'
}
}
}
}
}
}
The down side of this a approach is:
- Meta stage for the when block (or the same when block in every stage)
- every commit triggers a build even if there was no change in the subdirectory of the microservice
One possible solution would be to allow the when block at the toplevel and don't start a build when it is not matched.
