Details
-
Type:
Bug
-
Status: Resolved (View Workflow)
-
Priority:
Critical
-
Resolution: Duplicate
-
Labels:
-
Environment:Jenkins 2.19.1 on Ubuntu 14.04 x64
Windows 10 (1607) Connected via JNLP
-
Similar Issues:
Description
Freestyle jobs hang either at the beginning or the end of execution. I've let a few run for ~18 hours before they eventually crash with errors like these:
07:52:37 java.io.IOException: Unable to get hostname from slave. null
07:52:37 at hudson.plugins.perforce.PerforceSCM.checkout(PerforceSCM.java:1187)
07:52:37 at hudson.model.AbstractProject.checkout(AbstractProject.java:1278)
07:52:37 at hudson.model.AbstractBuild$AbstractBuildExecution.defaultCheckout(AbstractBuild.java:604)
07:52:37 at jenkins.scm.SCMCheckoutStrategy.checkout(SCMCheckoutStrategy.java:86)
07:52:37 at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:529)
07:52:37 at hudson.model.Run.execute(Run.java:1720)
07:52:37 at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:43)
07:52:37 at hudson.model.ResourceController.execute(ResourceController.java:98)
07:52:37 at hudson.model.Executor.run(Executor.java:404)
Reverting back to Jenkins 2.7.4 immediately resolved the issue. Issue was not observed with any Linux/OSX/older Windows nodes.
Attachments
Issue Links
- is related to
-
-
- Open
-
- links to
Activity
I believe the difference between my environment (where this hang causes all builds to stall) and your environment (where only that single node stalls) - is that we are using Pipeline builds, where you are using freestyle builds.
As part of the code for a pipeline build runs on the master, this hang seems to block all other activity on the master. That makes it a bit worse in the scenario where Pipeline is being used – but either way, its very bad.
We were able to recreate this problem with per-anniversary edition of Windows 10, so its not related to that.
It looks like this might be related to https://issues.jenkins-ci.org/browse/JENKINS-19445
I tried 2.32.3 with Oracle JDK8 today, issue persists.
It looks like upgrading my master to Ubuntu 16.04 resolved this issue. It was a fresh build, so it's possible Jenkins upgrades left some garbage in a config file somewhere, and my clean build didn't have that issue? Anyways, the issue is resolved for me, but it looks like people on certain configs/following certain upgrade paths may still have this issue.
That appears to be the case, all affected nodes are Win10 AE, but not all of my Win10 AE nodes are affected. It doesn't look like the affliction is consistent: Sometimes jobs will run fine on these nodes, sometimes they won't.