-
Bug
-
Resolution: Done
-
Major
-
None
-
None
-
None
Version: 0.0.6.Final
When running the same performance test as for standalone metrics on hawkular services the performance is ~40 times worst and it eventually fails because the WF jvm is stuck on garbage collecting.
I believe it's the same or similar issue which was reported against hawkular dist - https://issues.jboss.org/browse/JMAN4-102
Complete GC log is attached.
Heap dump is available here (1GB) - http://web.bc.jonqe.lab.eng.bos.redhat.com/haw_ser_dump.bin
hakular log is attached as well, but this log is from different run than run from which the heap dump and GC log are taken. This hawkular log is from run when 4GB of memory was used for WF.
When using 1GB of memory there is no exception in hawkular log but the test runs just 6 minutes so it's possible the exception would appear later.
I used 1GB to have heap dump with reasonable size but the issue should be visible there as well.
- causes
-
HAWKULAR-1114 Add shared ISPN configuration for metrics/alerts publishing workflow
- Closed