Publishing multiple test results from a jenkinsfile (or equivalent)

[Joshua Harlow]

Hi,

I am working through creating a jenkinsfile (and/or equivalent) for the openstack (nova) project we are using at godaddy,

The stages are broken up into (literally the following):

stage("Initiation")
flows = build()
node {
    start {
        net_info()
    }
}

stage("Checkout")
slack_stage(
    "Checking out (and stashing) repo for ${project}",
    {
        parallel(flows["cloners"])
    }
)

stage("Patch fetching")
slack_stage(
    "Checking out (internal) patch repo and stashing patches for ${project}",
    {
        parallel(flows['patch_getters'])
    }
)

stage("Unmodified (unit) testing")
slack_stage(
    "Unit testing unmodified ${project}",
    {
        parallel(flows["testers"])
    }
)

stage("Augmentation")
slack_stage(
    "Augmenting ${project}",
    {
        parallel(flows["augmenters"])
    }
)

stage("Modified (unit) testing")
slack_stage(
    "Unit testing modified ${project}",
    {
        parallel(flows["post_testers"])
    }
)

stage("Finished")
finish()

The unmodified testing there will publish before-patch unit test results and the post-patch modification will also publish another set of unit test results,

When this happens it seems like jenkins only shows the first and not both sets of unit test results; is there a plugin that would make both show up?

For recording I am using the following:

 step([$class: 'JUnitResultArchiver',
testResults: 'results.xml', allowEmptyResults: true])

Maybe there is something special people are doing in groovy to make it work,

Tips would be appreciated :)

-Josh

[Sean Flanigan]

I know it's weeks since this was asked, but I suppose it can't hurt to add some information. This problem is https://issues.jenkins-ci.org/browse/JENKINS-27395

I'm using this kludge inside Jenkinsfile (where 'appserver' is used to namespace the test results):

def integrationTests(def appserver) {

  sh "./mvnw verify"

  def testFiles = '**/target/failsafe-reports/TEST-*.xml'

  setJUnitPrefix(appserver, testFiles)

  junit testFiles

}

// from https://issues.jenkins-ci.org/browse/JENKINS-27395?focusedCommentId=256459&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-256459

def setJUnitPrefix(prefix, files) {

  // add prefix to qualified classname

  sh "shopt -s globstar && sed -i \"s/\\(<testcase .*classname=['\\\"]\\)\\([a-z]\\)/\\1${prefix.toUpperCase()}.\\2/g\" $files"

}

Naturally, a regex is a terrible way of parsing XML, but the idea of the sed script is to add the prefix to every classname attribute inside a testcase element (<testcase classname=""). I'm using failsafe's filenames here, but the general idea should work for any junit xml file, as long as classname is the first attribute in the testcase element, and on the same line. (The regex could be a little simpler, but I wanted the processing to be idempotent.)

As an example, the wildfly results for org.example.ExampleTest now appear in Jenkins as WILDFLY.org.example.ExampleTest, separated from JBOSSEAP.org.example.ExampleTest.

--
You received this message because you are subscribed to the Google Groups "Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/ba6d8e88-a84b-491f-946a-37cc6a57aeaa%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.