Minor We are using p4-plugin in freestyle jobs and the "Template" workspace behavior
Regularly (but not always) we are seeing client root is set to "null" rather than being updated to the correct value. I have not been able to work out why some jobs have this problem but others do not. It does not seem to affect the syncing of the jobs, but it is confusing for users.
It seems similar to an old issue that we had (possibly JENKINS-26589 )
We are also seeing the following error sometimes:
ERROR: P4 Task: failed: com.perforce.p4java.exception.P4JavaException: com.perforce.p4java.exception.P4JavaException: hudson.AbortException: P4JAVA: Error(s):
Path '/null/...' is not under client's root 'null'.
- is duplicated by
-
JENKINS-54511 When using the SpecWorkspaceImpl workspaces are update twice every build
-
- Closed
-
- is related to
-
-
- Closed
-
[JENKINS-52590] p4-plugin sets client root to null
Karl Wirth
added a comment - Have contacted Andrew directly via the support channel. Karl Wirth
added a comment - Thanks to Andrew we were able to find the cause. Andrew spotted that the null client times matched the time the poll was run.
Reproduction steps:
1 - Create a polled Freestyle job
2 - Build Job once.
3 - Wait till next poll (or use Poll Now plugin).
4 - Look at the polling log for the client name used in the polling. Run 'p4 client -o' on that client. The 'Root:' is 'null'.
Karl Wirth
added a comment - Thanks to Andrew we were able to find the cause. Andrew spotted that the null client times matched the time the poll was run.
Reproduction steps:
1 - Create a polled Freestyle job
2 - Build Job once.
3 - Wait till next poll (or use Poll Now plugin).
4 - Look at the polling log for the client name used in the polling. Run 'p4 client -o' on that client. The 'Root:' is 'null'.
Paul Allen
added a comment - Ready for release. Paul Allen
added a comment - Released in 1.9.6 Karl Wirth
added a comment - - edited JENKINS-59300 - Same bug but new one is for template workspaces. 
I can confirm this, but our setup is slightly different. We're using multibranch pipelines with Jenkinsfiles scanned over streams. I see this generally the first sync per pipeline, e.g. when creating the client. I've added a retry around the checkout step that "usually" takes care of the null ref (the second time generally works), but it not always. It looks like a timing issue to me, since it consistently (with an exception once or twice a month) works after that.