| After restarting jenkins, EC2 nodes fail to start with the following stacktrace: ERROR: Unexpected error in launching an agent. This is probably a bug in Jenkins java.lang.NullPointerException at java.lang.ProcessEnvironment.validateValue(ProcessEnvironment.java:120) at java.lang.ProcessEnvironment.access$400(ProcessEnvironment.java:61) at java.lang.ProcessEnvironment$Value.valueOf(ProcessEnvironment.java:203) at java.lang.ProcessEnvironment$StringEnvironment.put(ProcessEnvironment.java:243) at java.lang.ProcessEnvironment$StringEnvironment.put(ProcessEnvironment.java:221) at hudson.slaves.CommandLauncher.launch(CommandLauncher.java:130) at hudson.plugins.ec2.ssh.EC2UnixLauncher.launchScript(EC2UnixLauncher.java:209) at hudson.plugins.ec2.EC2ComputerLauncher.launch(EC2ComputerLauncher.java:48) at hudson.slaves.SlaveComputer$1.call(SlaveComputer.java:294) at jenkins.util.ContextResettingExecutorService$2.call(ContextResettingExecutorService.java:46) at jenkins.security.ImpersonatingExecutorService$2.call(ImpersonatingExecutorService.java:71) at java.util.concurrent.FutureTask.run(FutureTask.java:266) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) If we go to /configure and just save the current configuration, the nodes will successfully start after that. Since we have many nodes, this workaround is quite impractical. Same problem but different stacktrace: https://issues.jenkins-ci.org/browse/JENKINS-55473 |