atg crashed with tiger b32 using CMS collector if the heap of the server process is set to 12m.
at the end of "atgserver.log" there was the message
"[GC [ParNew (promotion failed): 3968K->3968K(4032K), 0.0223930 secs][CMS[root@jt"

How to reproduce the problem:
1. telnet to jtg-linux2.sfbay or a similiar machine
[root@jtg-linux2 client]# uname -a
Linux jtg-linux2 2.4.20-8smp #1 SMP Thu Mar 13 17:45:54 EST 2003 i686 i686 i386 GNU/Linux
2. export JAVA_HOME=<your java home>
3. export JAVA_ARGS="-XX:CMSInitiatingOccupancyFraction=10 -Xmx12m"
4. execute the script
/net/jtgb4u4c.sfbay/export/sail8/bigapps/runatg.ksh -client -XX:+UseConcMarkSweepGC

The problem happens from within a few minutes to a few hours.

So far I have no luck to reproduce the problem on solaris sparc.

###@###.### 2003-12-23

On solaris sparc, atg server OOM and then died if heap is set to 12m. I think
it's expected because of the small heap size.
I also searched in the log file on RedHat9, but did not find any message
indicating the atg server OOM.

###@###.### 2003-12-24