-
Bug
-
Resolution: Unresolved
-
Major
-
OS RedHat 6.7
jenkins 1.644
job-node-stalker-plugin 1.0.5
When using nodestalker plugin to ensure jobs are running on same workspace as their upstream job (and obviously on same node), I'm randomly facing issues where the job is sent to another node, where the expected workspace is not present.
In Jenkins system logs, I have plenty of these :
Apr 07, 2016 1:10:19 PM WARNING com.datalex.jenkins.plugins.nodestalker.wrapper.Util getProject
Cannot find job DeployerExample/forksl4-web-commun/release-1.12.0/01-forksl4-web-commun-release-1.12.0-init
Apr 07, 2016 1:10:19 PM WARNING com.datalex.jenkins.plugins.nodestalker.wrapper.MyNodeAssignmentAction setWorkspace
Job DRS-SL4/IHM/sl4-web-commun/release-1.0.2/01-sl4-web-commun-release-1.0.2-init not equal to DeployerExample/forksl4-web-commun/release-1.12.0/01-forksl4-web-commun-release-1.12.0-init. Custom workspace will not be set
Please note, that job this job does exist :
DeployerExample/forksl4-web-commun/release-1.12.0/01-forksl4-web-commun-release-1.12.0-init
And I expect the other job to macth this one properly (node+workspace).
By having a quick look at the code (MyNodeAssignmentAction.java / setWorkspace function), I suspect the findNearest function is (sometimes) confused here, as it doesn't find expected job/workspace.
Can this be the reason ?
Any idea to avoid this ?
I moved to another system instead based on Post-Build action: Trigger/Call builds on other projects.
There is indeed the parameter "Build on same node" that does the job properly.
Same principle is also applicable to Build others projects (manual step) for later manual launch.
This issue can be closed.