-
Bug
-
Resolution: Cannot Reproduce
-
Minor
-
None
-
Jenkins version 1.540 running in a standalone GlassFish v4.0_b89 server running JDK 1.7.0_45 on CentOS v5.8.
After changing the state of the `Enabled` checkbox for a plugin on the `Manage Plugins` page (either enabling or disabling) and using the button that appears at the bottom of the page to trigger a restart of the Jenkins application, we've noticed that the subsequent behavior of the Jenkins host JVM heap is a constant flatline. This is very much different than the standard garbage collection `sawtooth` we see after a restart of the Jenkins host JVM. It's almost like restarting the Jenkins application via that button on the plugins page fails to restart everything in the app properly. This results in an eventual OOM error after some time in our experience. This behavior can be seen by running VisualVM on the Jenkins host JVM and watching the heap behavior before and after the Jenkins restart.