When working with jenkins 2 (declarative) pipelines and maven I always have a problem with how to organize things within the pipeline to make it resusable and flexible.
On the one side I would like to seperate the pipepline into logical stages like:
pipeline
{
stages
{
stage('Clean') {}
stage('Build') {}
stage('Test') {}
stage('Sanity check') {}
stage('Documentation') {}
stage('Deploy - Test') {}
stage('Selenium tests') {}
stage('Deploy - Production') {}
stage('Deliver') {}
}
}
On the other hand I have maven which runs with
mvn clean deploy site
Simply I could split up maven to
mvn clean
mvn deploy
mvn site
But the 'deploy' includes all lifecycle phases from
- validate
- compile
- test
- package
- verify
- install
- deploy
So I saw a lot of pipline examples which do things like
sh 'mvn clean compile'
and
sh 'mvn test'
which results in repeating the validate and compile step a second time and waste "time/resources" in this way. This could be resolved with doing a
sh 'mvn surefire:test'
instead of running the whole lifecycle again.
So my question is - which is the best way to get a good balance between the jenkins pipline stages and the maven lifecycle? For me I see two ways:
- Split up the maven lifecycles to as much pipeline stages as possible - which will result in better jenkins user feedback (see which stage fails etc.)
- Let maven do everything and use the jenkins pipeline only to work with the results of maven (i.e. analyzing unit test results etc.)
Or did I missunderstand something in the CI/CD practice?