OK, I bumped into this issue recently, and I have a guess of the cause: I think it's corrupt slave.jar files. I had two nodes running two boxes: jenkinsdev@box1 and jenkinsdev@box2. Both slaves were configured to write to /home/jenkinsdev/jenkins_workspace. This home directory is mirrored between box1 and box2. As a result, both slaves try to write to the same file, which is /home/jenkinsdev/jenkins_workspace/slave.jar. This would explain why the traceback complains about the zip file. I couldn't find anything obviously corrupt in my slave.jar file, but I didn't look very hard. This would explain why restarting the master causes this. This restarts all the slaves, so they start up simultaneously, and try to write the slave.jar simultaneously. |