Greetings.
We have two RedHat Linux servers, Dlv1 and Dlv2, running:
Linux 2.6.18-274.3.1.el5 #1 SMP Fri Aug 26 18:49:02 EDT 2011 x86_64 x86_64
x86_64 GNU/Linux
Each has an identical 7.0.23 Tomcat engine running, front-ended by
Apache/2.2.3. Each Tomcat runs exact binary-compatible applications. (Dlv2 is
running Dlv1's image.)
Dlv2 is leaking heap memory and is consistently slower. A heapdump analysis
of Dlv2 shows finalizers queuing up on Dlv2 pointing to a possible backlog into
the backend database. Threads, monitors etc. look healthy on Dlv2. Dlv2 is
out of the load-balancer pool, needless to say. Database queries directly from
Dlv1 and Dlv2 show no differences.
What else could cause JVM runtime performance differences?
An md5sum comparison of the "/usr/lib/jvm/java-1.6.0-sun-1.6.0.30.x86_64/"
installation on Dlv1 and Dlv2 show differences in several files. But these
seem to be a repeatable difference almost like a watermark, a signature or the
IP address encoded for a machine-specific difference.
I was going to tar up Dlv1's Java1.6 install and untar it on DLv2. But I
notice md5sum differences in the older JDK1.5 installs as well. Are these md5
checksum differences common across RPM installs of Java?
Do you have any tips on other causes that might impact a JVM's runtime
operation?
Any suggestions on what else I might try?
Thanks!
-Shanti
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
