-
Bug
-
Resolution: Fixed
-
Critical
-
Jenkins 2.32.5, plugin version 1.0.3 Ubuntu 20 GitLab plugin
Relevant bit of my Jenkins file:
stage('Checkout') { steps { cleanWs() checkout scm scmSkip(deleteBuild: true, skipPattern:'.*skip-ci.*') } }
console output show message "SCM Skip - build skipped" but the build never stops. This results in a never ending cycle of builds
Note: I am using multibranch pipelines.
- is duplicated by
-
-
- Resolved
-
- is related to
-
-
- Resolved
-
-
-
- Closed
-
[JENKINS-67436] SCM Skip Plugin does not abort pipeline
Gabriella
added a comment - An update on this issue:
1) I cannot seem to change the pattern in the Configure System interface. It never saves the change.
2) Using the "default" ci skip results in one extra build happening. The pattern is recognized, but the build does not stop and therefore pushes again, triggering another build. Seems to stops after this second build. Better than before but not ideal.
Please fix ASAP!
Gabriella
added a comment - An update on this issue:
1) I cannot seem to change the pattern in the Configure System interface. It never saves the change.
2) Using the "default" ci skip results in one extra build happening. The pattern is recognized, but the build does not stop and therefore pushes again, triggering another build. Seems to stops after this second build. Better than before but not ideal.
Please fix ASAP! gturek if changes are not being saving from the "Configure System" page, you may have one or more plugins installed that have issues with the configuration form modernization that was implemented in Jenkins 2.277.1.
See https://community.jenkins.io/t/migrating-jenkins/894/5 for detailed instructions that will allow you to investigate your specific system to see if there are plugins causing the problem.
Gabriella
added a comment - Hi Mark, no I don't have any issues saving the "Configure System" page edits except for the SCM Skip plugin's config (changing the default regex). It won't stick
Gabriella
added a comment - Hi Mark, no I don't have any issues saving the "Configure System" page edits except for the SCM Skip plugin's config (changing the default regex). It won't stick 
Florian Fetz
added a comment - I had a similar issue with Multibranch Pipelines. For some pipelines, the build would always stop and for some it would never do that, even though scmSkip reported it wanted to stop the build.
What worked for me was to put a "sleep(1)" after the "scmSkip()" call. It's not pretty but it obviously does the job.
Florian Fetz
added a comment - I had a similar issue with Multibranch Pipelines. For some pipelines, the build would always stop and for some it would never do that, even though scmSkip reported it wanted to stop the build.
What worked for me was to put a "sleep(1)" after the "scmSkip()" call. It's not pretty but it obviously does the job. 
Axel Müller
added a comment - I checked out the code of the plugin but had to introduce the Jenkins BOM and use a more recent Jenkins version in order to be able to build the plugin from source (created a PR for this: https://github.com/jenkinsci/scmskip-plugin/pull/8)
Without the video tutorials "Modernizing Jenkins Plugins" by markewaite I would not have been able to compile and run it. Thank's a lot for them!
For various tests I created a sandbox project which was used by Jekins runing inside the IDE (using: `hpi:run`).
The SCMSkip plugin accepts a `deleteBuild` parameter which deletes the build successfully. However, I'm not sure if this approach is valid at all since by doing so it might also remove the information from Jenkins that this changeset has already been processed. At least this is my explanation why another build is scheduled by Jenkins which should not happen.
Since I don't mind if the build is not deleted if it is skipped because of commit message content I did some tests with `deleteBuild: false`. With this configuration build were skipped correctly and not scheduled again. If a commit message without skip directive comes afterwards the build is processed normally. Nothing to complain about - so will go with this configuration for now ...
Axel Müller
added a comment - I checked out the code of the plugin but had to introduce the Jenkins BOM and use a more recent Jenkins version in order to be able to build the plugin from source (created a PR for this: https://github.com/jenkinsci/scmskip-plugin/pull/8)
Without the video tutorials "Modernizing Jenkins Plugins" by markewaite I would not have been able to compile and run it. Thank's a lot for them!
For various tests I created a sandbox project which was used by Jekins runing inside the IDE (using: `hpi:run`).
The SCMSkip plugin accepts a `deleteBuild` parameter which deletes the build successfully. However, I'm not sure if this approach is valid at all since by doing so it might also remove the information from Jenkins that this changeset has already been processed. At least this is my explanation why another build is scheduled by Jenkins which should not happen.
Since I don't mind if the build is not deleted if it is skipped because of commit message content I did some tests with `deleteBuild: false`. With this configuration build were skipped correctly and not scheduled again. If a commit message without skip directive comes afterwards the build is processed normally. Nothing to complain about - so will go with this configuration for now ...
Original issue fixed in https://plugins.jenkins.io/scmskip/releases/#version_50.vfb_3a_f04242a_a_
If any the configuration-saving issue still exists, please create a new ticket.
I am thinking this is perhaps a problem with the gitlab plugin as the scmSkip plugin has not been updated in a while, but the Gitlab plugin has