> From: Ken Bowen [mailto:kbo...@als.com] > Subject: GC behavior elucidation > > My gclog shows routine GCs approx every 3-4 mins that look like this: > 2010-03-17T01:28:05.888-0400: 543717.992: > [GC 543717.992: > [DefNew: 6222K->487K(6528K), 0.0030010 secs] > 44060K->38706K(64832K), 0.0030920 secs] > [Times: user=0.01 sys=0.00, real=0.01 secs]
Looks reasonable. Make sure the contracted number stay about where they are; if they keep increasing, you've still got a leak. > Since 3/10, I've seen 5 Full GCs > 2010-03-17T01:07:34.278-0400: 542486.382: > [Full GC 542486.382: > [Tenured: 58226K->35935K(58304K), 0.1316550 secs] > 64221K->35935K(64832K), > [Perm : 21200K->21200K(21248K)], > 0.1317440 secs] > [Times: user=0.13 sys=0.00, real=0.13 secs] > > At the same time, catalina.out shows a lot of entries like this: > INFO | jvm 1 | 2010/03/17 01:07:34 | [Unloading class > sun.reflect.GeneratedMethodAccessor233] Normal for a full GC - unused classes are being discarded. > I've seen 7 GCs of the third type: > 2010-03-23T20:44:55.835-0400: 1131527.939: > [GC 1131527.939: > [DefNew (promotion failed): 6205K->5999K(6528K), 0.0047690 secs] > 1131527.944: > [Tenured: 58304K->36397K(58304K), 0.1396970 secs] > 64399K->36397K(64832K), > [Perm : 21631K->21631K(21760K)], > 0.1446100 secs] > [Times: user=0.14 sys=0.00, real=0.14 secs] > I assume that the third type of GC is something intermediate between > the routine (every 4 mins) small collection and a Full GC. No, it's a minor collection that turned into a major one due to the tenured space being full. If you haven't prevented internal heap component resizing via command-line settings, the JVM will try to adjust the sizes of the spaces to try to avoid reoccurrence. However, your tenured usage seems to be creeping up, so you may well still have a leak. - Chuck THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
