Publish Server Memory Leakage Issue and Details

[radhikak8817019]

Hi All,

We faced issues with the publishers, one publish server went down due to ”java.lang.OutOfMemoryError: Java heap space” and the second publisher is also throwing Out of Memory exception.

When we had this problem a few months back it got resolved when we decided to increase the JVM heap size on the Publishers.
But there still seems to be a memory leak causing the heap space to OOM.

The increasing JVM didn’t actually solve the root-cause since we raised the “roof” for a while and we seem to be hitting it again…we need to investigate on what is causing the JVM to OOM and get a better understanding on why it occurs.

On further analysis we found : GC for JVM has thrown an error and it’s not the same GC that is run for CQ repository, so both are completely different. 

Error Details

-------------------------
30.04.2015 00:00:14.382 *ERROR* [FelixDispatchQueue] com.day.crx.sling.crx-auth-token FrameworkEvent ERROR (org.osgi.framework.ServiceException: Service factory exception: javax/servlet/Servlet) org.osgi.framework.ServiceException: Service factory exception: javax/servlet/Servlet

Kindly pour in your suggestions in order to resolve this issue.

Thanks,

RK

[Jörg_Hoh · Accepted solution]

Hi,

The exception you posted might indicate a different problem, but not a memory leak. Can you configure your JVM to create a heapdump in case it runs out of memory.

Note: the repository itself has a number of cleanup jobs, with the "Datastore Garbage Collection" being one of them. These jobs are totally unrelated to the the Garbage Collection of the JVM, the only common aspect of these is the name.

kind regards,
Jörg