Dear Leon,
i'm performing a jmap -histo:live on a tomcat which froze because of
full (8GB) old space gen, and my top 10 looks like:
num #instances #bytes class name
----------------------------------------------
1: 19268009 2655683368 [C
2: 17410092 1392807360
org.apache.tomcat.util.buf.MessageBytes
3: 17411294 1114322816
org.apache.tomcat.util.buf.ByteChunk
4: 17410562 974991472
org.apache.tomcat.util.buf.CharChunk
5: 19275640 771025600 java.lang.String
6: 8697280 626204160 javax.servlet.http.Cookie
7: 8699584 417580032
org.apache.tomcat.util.http.MimeHeaderField
8: 188 134247456
[Lorg.apache.tomcat.util.http.MimeHeaderField;
9: 198645 114493776 [Ljava.lang.Object;
10: 21303 30450504 [B
So how did I manage to get almost 9 million of mimeheaderfields and
cookies and 17 millions message bytes and chunks.
During normal operation mode the number of the above doesn't increase
except before the gc freeze.
You can try to generate a heap dump and analyse it using jhat. Trouble
is, jhat requires insane amounts of memory. So for you 8GB dump you'd
need something in the order of 32GB RAM for jhat.
If that is feasible, you could lower the memory limits on your
machine, run the tests again to get a smaller dump with (hopefully)
the same characteristics.
--
Kees Jan
http://java-monitor.com/forum/
kjkos...@kjkoster.org
06-51838192
The secret of success lies in the stability of the goal. -- Benjamin
Disraeli
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
