Integration

Can someone tell if it is likely that TLA bailout errors are just due to too large of a heap or if there are other things we need to look at? We plan to decrease the heap to 1.75 GB and maybe try the -Xnoopt flag. The server has 4 GB RAM (32-bit OS) with /3GB switch set and 2GB of heap (WL10 with updated beasvc wrapper to allow over 2GB).

|memory ||Wed Jan 13 04:35:05 2010||04680| 44424.157-44424.767: GC 2097152K->596067K (2097152K), 609.528 ms
|memory ||Wed Jan 13 04:35:06 2010||04680| TLA bailout requested (heap=9C4701F0)!
|memory ||Wed Jan 13 04:35:06 2010||04680| TLA unwind thread links.
|memory ||Wed Jan 13 04:35:06 2010||04680| Throwing OutOfMemory: CG(q0) |sun/util/resources/TimeZoneNames.getContents()||Ljava/lang/Object;| JVM@cgFail (src/jvm/code/codemanager.c:692). Java h
|memory ||Wed Jan 13 04:35:07 2010||04680| 44425.753: parallel nursery GC 1466732K->666476K (2097152K), 35.034 ms
|memory ||Wed Jan 13 04:35:07 2010||04680| 44426.708: parallel nursery GC 1531980K->930124K (2097152K), 35.485 ms
|memory ||Wed Jan 13 04:35:08 2010||04680| 44427.401: parallel nursery GC 1531724K->952602K (2097152K), 38.274 ms
|memory ||Wed Jan 13 04:35:09 2010||04680| 44427.999: parallel nursery GC 1607656K->1005864K (2097152K), 33.503 ms

|memory ||Wed Jan 13 05:16:42 2010||04680| 46920.947: parallel nursery GC 2097015K->2097152K (2097152K), 60.256 ms
|memory ||Wed Jan 13 05:16:42 2010||04680| 46921.008-46921.619: GC 2097152K->511710K (2097152K), 611.478 ms
|memory ||Wed Jan 13 05:16:42 2010||04680| Allocation of 320760 bytes failed for heap "<unnamed>"
|memory ||Wed Jan 13 05:16:42 2010||04680| TLA bailout requested (heap=9C4701F0)!
|memory ||Wed Jan 13 05:16:42 2010||04680| TLA unwind thread links.
|memory ||Wed Jan 13 05:16:44 2010||04680| 46922.760: parallel nursery GC 1366377K->785449K (2097152K), 47.578 ms

Would monitoring the non-heap memory usage mxbean tell me anything about whether I'm going to get a TLA bailout error?

Thanks,
-Bill

Edited by: Bill S. on Jan 13, 2010 11:06 AM