Serious performance issue using Oracle 9.2.0.4 and RAC/TAF

[bobneville]

Hi

We are using a third party package that uses JBOSS 3.2.2. Originally the application connected to an Oracle instance (9.2.0.4 64 bit) running on a single server and performance was acceptable but not great.
We have now moved the database to it's real server which is a dual node
Solaris 8 cluster using Oracle RAC and TAF - again 64 bit 9.2.0.4.

Database performance is great, less than .1 second response time, however the application servers running JBOSS are maxed out after a short period of time - 99% CPU etc etc. The only way we can resolve the issue is to reboot the app servers. We also notice a very large number of hanging threads which have been dumped and analysed but give us no clues.

All other NON JBOSS applications that use this set up perform very well !!

Is there anything we should have done to allow JBOSS to perform against this cluster correctly or, more importantly, will this type of set up work ???

Thanks in advance

Bob

[mathewdenis]

This is due to less Internet bandwidth of ISP. Increase the Internet bandwidth to avoid this problem. If Internet bandwidth goes down all the request gets qued up and eat all the virtual memory of the server.

[bie-developer]

Um, are you sure you have enough information to provide an answer? The original post said nothing about ISPs or bandwith, and he would be seeing java.lang.OutOfMemory errors if he had a memory problem.

[bobneville]

We have fixed the problem !!

It has nothing to do with bandwidth, there is no ISP involved as this is an internal company system.

The problem was that we had the wrong version of the Oracle OCI driver loaded. Once the correct version was installed the problem resolved itself.

The reason why the performance was so bad against the cluster would appear to be linked to one of the patches in the OCI Oracle drivers