We are migrating our Blue Martini Commerce application (only supported on WebLogic 10.3.0) to a new host (Redhat 6.3 on a VMWare ESX vm). We are seeing extremely slow start up times for our managed server(s) that is basically 20x slower than our current production.
As a for instance the Publish managed server takes ~30 - 45 seconds in current production and in the new environment it takes ~10 minutes.
The setup uses the same domain structure and JVM as the current production environment. The same setup files are used. We use jdk1.6.0_33 on 64 bit architecture. We used the generic 64bit weblogic installer and used pack / unpack utilities to migrate the domain.
The JAVA_OPTS to start this server are: "-d64 -Xms256m -Xmx512m -XX:PermSize=48m -XX:MaxPermSize=256m"
The sysadmins have checked /etc/sysctl.conf and /etc/limits.conf to ensure we were not hitting some kind of process limit. As I am not sure what this managed server does from a Blue Martini perspective during the phase of startup I also had the DBA check to ensure that Oracle RAC (18.104.22.168) wasn't also hitting some kind of process limit or if there was a tns listener issue.
The new host is quite a bit stricter with their server lock downs so there are a few differences....
- Redhat 6.3 in new env, RH 5.7 in current
- SElinux is targeted in new env and disabled in current
- VM in new env and dedicated hardware in current
- iptables disabled in current. It was enabled in new prod but I had them disable it just in case
I apologize for not being more specific. I am mostly hoping got some tips. I do not have the typical root access I would normally have in this environment. I am just hoping got a path forward. I did a few 'kill -3' to see if there are blocked threads and I got nadda. The service works for all intents and purposes it is just painfully slow.
Thanks you all in advance for reading and best regards. Wade