Hello, we have been observing that one (or at times two) of our Weblogic 10 instances are consuming maximum CPU (>90%). Every week-end we restart the servers, so at that time it doesn't happen, but then during busy hours it starts coming up again & again. It's 2 CPU machine with two instances running. Using the top command we have identified that ‘OC Main Thread’ is consuming maximum CPU. In the thread-dump, only the following information has been mentioned:
+"(OC Main Thread)" id=3 idx=0xc tid=25472 prio=5 alive, in native, daemon”+
Can someone please tell us what could be the possible reasons for which this Weblogic/JRockit thread "(OC Main Thread)" would consume maximum CPU?
Edited by: 952315 on Aug 12, 2012 1:48 PM
Edited by: 952315 on Aug 12, 2012 1:53 PM
We have been still analyzing this problem - the main problem seems to be frequent garbage collection in 'Old Space' - we had the JVM setting as -XXtlasize:min=8k,preferred=64k
-XXlargeobjectlimit:8k - so larger objects were getting copied directly into the 'Old Space' instead of 'Nursery Space'. Oracle has recommended to install a JRockit patch (Doc ID 1209624.1) and change the JVM setting. We need to try those things out in our Production environment.