How to use logRotator when parallel jobs produce artefacts? - jenkins

Question

My pipeline is compiling on a windows and linux machine in parallel.
Since I used the parallel directive logrotators does not work anymore and I don't manage to find what is wrong
All artefacts are keeping stored .
here is a sample of my Jenkinsfile
properties([gitLabConnection('numagit'),
buildDiscarder(
logRotator(
numToKeepStr:'1', artifactNumToKeepStr:'1'
)
)
]
)
parallel (
'linux' : {
node('linux64') {
stage('Checkout sources') {
echo 'Checkout..'
checkout(scm)
}
gitlabBuilds(builds:[
"Compiling linux64"
] ) {
try {
stage('Compiling linux64') {
gitlabCommitStatus("Compiling linux64") {
sh('rm -rf build64')
sh('mkdir -p build64')
dir('build64')
{
......
}
archiveArtifacts(artifacts: 'build64/TARGET/numalliance/MAJ/data.tgz', fingerprint: true)
archiveArtifacts(artifacts: 'build64/TARGET/numalliance/MAJ/machine.sh', fingerprint: true)
}
}
} catch (e) {
currentBuild.result = "FAILURE" // make sure other exceptions are recorded as failure too
// en cas d erreur on archive la sortie du CMake
archiveArtifacts("build64/CMakeFiles/CMakeOutput.log")
}
}
cleanWs()
}
},
'windows' : {
// Noeud de compilation windows
node('win32') {
def revision = ""
stage('Checkout sources') {
echo 'Checkout..'
checkout(scm)
}
try {
stage('Compilation windows') {
gitlabCommitStatus("Compilation windows") {
echo 'Building win32 version'
....
}
}
stage('Packaging for win32') {
gitlabCommitStatus('Packaging for win32') {
....
dir('win32/TARGET/numalliance/MACHINE'){
...
archiveArtifacts(artifacts: '*.exe', fingerprint: true)
}
}
}
} catch (e) {
currentBuild.result = "FAILURE" // make sure other exceptions are recorded as failure too
}
cleanWs()
}
}
)

Answer 1

The properties function changes the configuration of the job. It is the same as if you were to go to the configuration page of the job and change the log rotator settings manually. It seems you want to achieve that 5 development builds and one of everything else are kept but what actually happens is that the pipeline will keep 5 of anything when a develop job has run last and it will keep 1 when something else ran last.
The easiest fix would be to use seperate jobs (with the same pipeline code). In that case you would have one job that only keeps the last build and another (the develop job) which keeps the last 5.