Links may not function; however, this content may be relevant to outdated versions of the product.
Troubleshooting: JVM hangs on conclusion cache read or update (HFix-10223 replaces HFix-8969)
Systems with HFix-8969 installed (either as a primary hotfix or as a dependent hotfix) experience JVM failure. Production application server JVM is unresponsive, causing the user sessions to freeze. This occurs on all supported application servers and JVMs.
To address this issue, HFix-10223 replaces HFix-8969.
A Java Core dump thread analysis shows that most of the Web container user threads are in a WAITING (parking) state, executing either a call stack containing ConclusionCacheImpl.noteChange() or ConclusionCacheImpl.getOrCreate() methods. These methods wait indefinitely trying to get a ReentrantReadWriteLock$ReadLock or ReentrantReadWriteLock$WriteLock by means of the wrapper PRPC class ReadWriteLockUpgrade. As a concurrency issue, the problem is difficult to reproduce.
Following are example thread call stacks showing the parked state — java.lang.Thread.State: WAITING (parking) — in a thread dump generated after the JVM freeze.
HFix-10223 corrects and replaces HFix-8969. HFix-10223 prevents the upgrading of locks from Read to Write, which caused the unresponsive application server JVM, and it now synchronizes conclusion cache updates.
If you experience the problem described in this article, follow these steps:
- Scan your PRPC 6.3 SP1 system to determine whether HFix-8969 is installed.
From the button, click > > > .
- If HFix-8969 is installed, go to PDN My Support Portal, Quick Actions, and click Create Support Request.
- In your Support Request (SR), ask for HFix-10223.
- Install HFix-10223 from the DL file you receive from GCS.
Previous topic Troubleshooting: CA Wily Introscope memory leak when SMA connected to remote PRPC servers Next topic Troubleshooting: Unexpected large numbers of queries to pr_engineclasses table causes unnecessary DB traffic (HFix-4114)