-
Bug
-
Resolution: Fixed
-
Major
-
None
-
Fedora 16 master and RHEL5 slave
I enabled Xvfb in a job, configured both the master and the slave to create a default Xvfb config and properly locate the binary on the slave. Every time I run the job the result is the following backtrace:
Xvfb startingFATAL: null
java.lang.NullPointerException
at com.cloudbees.jenkins.plugins.customtools.CustomToolInstallWrapper$2.toEnvVars(CustomToolInstallWrapper.java:142)
at com.cloudbees.jenkins.plugins.customtools.CustomToolInstallWrapper$2.launch(CustomToolInstallWrapper.java:131)
at hudson.Launcher$ProcStarter.start(Launcher.java:338)
at org.jenkinsci.plugins.xvfb.XvfbBuildWrapper.decorateLauncher(XvfbBuildWrapper.java:191)
at hudson.model.AbstractBuild$AbstractBuildExecution.createLauncher(AbstractBuild.java:549)
at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:475)
at hudson.model.Run.execute(Run.java:1502)
at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:46)
at hudson.model.ResourceController.execute(ResourceController.java:88)
at hudson.model.Executor.run(Executor.java:236)
[JENKINS-14790] NullPointerException starting Xvfb on remote slave
Code changed in jenkins
User: Zoran Regvart
Path:
src/main/java/org/jenkinsci/plugins/xvfb/XvfbBuildWrapper.java
src/main/resources/org/jenkinsci/plugins/xvfb/Messages.properties
http://jenkins-ci.org/commit/xvfb-plugin/40e566d37f2cafb78b659693fcc416d79fe7dfa6
Log:
JENKINS-14790 NullPointerException starting Xvfb on remote slave
zregvart
added a comment - Yes I found a similar issue when Xvfb plugin is enabled on the job before tool installation/configuration in the jenkins (master/slave) configuration. You seem to have customtools-plugin installed which clouded this issue. And I'm unable to reproduce your exact issue with customtools-plugin installed (environment having null value).
Unfortunately the Xvfb plugin, currently at least, does not offer automatic tool installation, and you have to install Xvfb on your master/slave machines and configure the path in the jenkins configuration.
I've made a small fix so that now if no installation is configured in jenkins a warning message is placed in the console of the build.
Is it alright with you that I close this issue?
zregvart
added a comment - Yes I found a similar issue when Xvfb plugin is enabled on the job before tool installation/configuration in the jenkins (master/slave) configuration. You seem to have customtools-plugin installed which clouded this issue. And I'm unable to reproduce your exact issue with customtools-plugin installed (environment having null value).
Unfortunately the Xvfb plugin, currently at least, does not offer automatic tool installation, and you have to install Xvfb on your master/slave machines and configure the path in the jenkins configuration.
I've made a small fix so that now if no installation is configured in jenkins a warning message is placed in the console of the build.
Is it alright with you that I close this issue? 
Brian Sipos
added a comment - My failing setup defined a default path for Xvfb, but did not define the same path for each slave (all the slaves would have path /usr/bin) so I expected that the default path would be used.
At least having an error message with instructions (as you have changed) instead of a generic null exception is definite improvement. I am okay with closing the issue as-is.
Brian Sipos
added a comment - My failing setup defined a default path for Xvfb, but did not define the same path for each slave (all the slaves would have path /usr/bin) so I expected that the default path would be used.
At least having an error message with instructions (as you have changed) instead of a generic null exception is definite improvement. I am okay with closing the issue as-is. 
Hum actually can't reproduce on centos5:
Name: Xvfb
Directory in which to find Xvfb executable: '/usr/bin/'
Slave was able to start without NPE