How to solve problem After patch 10.2.0.4 applying on owb 10.2.0.1

Question

HI OWB Gurus,
We have recently a patchapplied for oracle 10g .
prev version was oracle 10 g r2 10.2.0.1
new version is 10.2.0.4 ( patchset 4)
After patch applied at database and owb level there are some problem we are facing like
ETL process is very slow , cpu is going to 100% after start any mapping.
and audit browser are not opening.
Can any one help me on this so that we can resolve the problem.
- Umesh

Answer 1

Hi Umesh,
This might help you in diagnosing the client's performance
Accurate optimizer statistics are generally very important. They are gathered automatically at OWB install time and the Design Center -> Tools -> Optimize Repository menu selection is available to refresh them on an as needed basis.
The DBA can always help by pre-expanding tablespaces, making sure multiple tablespaces are available on separate disks or disk arrays so that I/O to indexes, metadata tables, snapshot tables, DB undo and redo logs, etc., can be evenly distributed.
With more complex metadata in this OWB release, you can expect many JDBC round-trips to the DB. Network latency can have a big impact on performance. To see how much time the owbclient uses for JDBC roundtrips to the DB, add the following line to Preference.properties in <owb home>/owb/bin/admin on your client machine:ClientProfile=true
Run OWB, then check ClientProfile.log in the admin directory. Keep in mind that a ping to localhost is probably about 0.025 ms, to another machine in the LAN about 0.25 ms. If the roundtrips are greater than 40 ms, then the issue is likely related network latency.
The two basic workarounds for network latency are:
-Remote control an OWB client machine that is physically close to the server.
-Do design-time work on a local repository, then MDL export the work for import to a central repository.
If the client and server are not on the same (high bandwidth) LAN, then total available bandwidth may become a factor. Same thing if the DB server is heavily loaded with respect to CPU or I/O, or if all your OWB design-time metadata is located on the same tablespace on the same hard drive. Distributing projects across multiple repositories which use their own
tablespaces spread across multiple hard drives would be very desirable.
The object editors are more functional and complex, as such, the required processing power to render them is greater (processor > 2GHz) Also note the OWB client software is mostly single-threaded, so it won't take advantage of multiple processors on one machine.
Long running sessions or large bulk operations (MDL import, validation, generation, deployment over large or multiple modules) can push up against the JVM memory limit. If performance begins to suffer due to memory, either:
a)exit and restart the OWB session;
b) if adequate physical memory exists, increase the -Xmx and -Dlimit values in the owbclient and OMBPlus startup scripts to better accommodate large operations.
ie.
If you have 2 GB or more, backup and edit owb\bin\win32\owbclient.bat file and modify line starting with %JAVAPATH
ie.
%JAVAPATH%\jre\bin\java -Xms128M -Xmx1024M -Dlimit=1024M -XX:MaxPermSize=1024M
This will add more memory to JVM for OWB client. Do the same for owb.cl file
Hope this will add some value.
Best Regards,
Prasad Bolla.