Search code examples
mavenjenkinspom.xmlnexusartifacts

What is the best practice for archiving Jenkins pipeline artifacts

I see that you can add a goal to a maven Pom to archive artifacts to a repo manager such as nexus

https://www.baeldung.com/maven-deploy-nexus

distributionManagement>
   <snapshotRepository>
      <id>nexus-snapshots</id>
      <url>http://localhost:8081/nexus/content/repositories/snapshots</url>
   </snapshotRepository>
</distributionManagement>

But Jenkins also allows you to do this from the pipeline itself

https://wiki.jenkins.io/plugins/servlet/mobile?contentId=96076078#content/view/96076078

freeStyleJob('NexusArtifactUploaderJob') {
    steps {
      nexusArtifactUploader {
        nexusVersion('nexus2')
        protocol('http')
        nexusUrl('localhost:8080/nexus')
        groupId('sp.sd')
        version('2.4')
        repository('NexusArtifactUploader')
        credentialsId('44620c50-1589-4617-a677-7563985e46e1')
        artifact {
            artifactId('nexus-artifact-uploader')
            type('jar')
            classifier('debug')
            file('nexus-artifact-uploader.jar')
        }
        artifact {
            artifactId('nexus-artifact-uploader')
            type('hpi')
            classifier('debug')
            file('nexus-artifact-uploader.hpi')
        }
      }
    }
}

What is the difference between the two approaches, is one more commonly used?

Solution

  • The main difference is that if you use maven, you can manually add artifacts to Nexus from your local computer, using mvn deploy. So it really comes down to how you want to create the artifacts that end up being used in production. In my experience, the preferred way of doing this is through using Jenkins. The advantage of using Jenkins is that you link your new version builds to other activities, as well as the possibility to trigger a release when certain conditions have been met, rather than starting the build manually. Also, you end up with all versions being built on the same platform, and you avoid differences between computers if every developer builds such versions from their own computer.

    But you might still need the maven configuration. Jenkins might use this information to find the URL to upload the artifact to (your example says nothing about how Jenkins finds Nexus), and sometimes it is useful to upload a SNAPSHOT-version, or some other temporary version not meant for production. In your example you only define Nexus for upload SNAPSHOT-versions, I guess this is done on purpose to enforce a rule that uploading final version from local computers is disallowed.

    By the way, having a repository defined in your pom.xml does not automatically mean that anything will be uploaded. It is only if you do mvn deploy with a repository defined in your pom that something will be uploaded.