[
https://issues.jboss.org/browse/ISPN-4498?page=com.atlassian.jira.plugin....
]
William Burns commented on ISPN-4498:
-------------------------------------
Also a side note that updating to log4j2 would also solve this as this is only an issue
with Log4j1.2. Even after the JBLOGGING-106 is fixed we will still need to call to clear
the NDC by hand with log4j1.2, log4j2 uses a java.lang.ThreadLocal so the leakage
wouldn't occur even without the clear.
OutOfMemoryError when CI is ran with tracing
--------------------------------------------
Key: ISPN-4498
URL:
https://issues.jboss.org/browse/ISPN-4498
Project: Infinispan
Issue Type: Bug
Security Level: Public(Everyone can see)
Components: Test Suite - Server
Affects Versions: 7.0.0.Alpha4
Reporter: William Burns
Assignee: William Burns
The CI will fail very often when tracing is enabled with some sort of OOM error.
http://ci.infinispan.org/viewType.html?buildTypeId=Infinispan_MasterHotsp...
Looking closer this appears to be an issue with an accumulation of Threads. These
threads aren't running either but rather are held in memory unneedingly. It appears
that they are being retained in the log4j1.2 NDC class in it's hashtable. Upon
further investigation we never call to NDC.remove which would clear up the current any
dead threads.
I have crated JBLOGGING-106 to fix this as well. In the mean time we shouldn't use
NDC.
--
This message was sent by Atlassian JIRA
(v6.2.6#6264)