OAS JVM restarting at random

We have had 2 instances this week of an OAS JVM restarting without any warning or log file entries. We have 2 servers, each with 2 JVM's, and only 1 JVM on each machine restarted (at different times on different days). Has anyone else seen this type of behavior, and, better yet, found out what could be causing this?

Not so much as a OutofMemory message in the log anytime before it restarted?

As Jeremy said check outofmemory in opmn log files.
Many times when this problem happens there are problems with export/import Excel, or gotoend grid on customized application with fetch or other Table IO on event Like "grid record is fetched" or "Write Grid Line before"

Luigi,

Thanks for your reply. I hadn't checked the opmn log files, and I did indeed see some out of memory errors (java heap space) -- I've included a small snippet below (I don't know if the GZIP relates to the restart, since it occurred 8 minutes before). With this showing up afterwards as a result of 1 or more offending programs, how can you trace it back to the program(s) at fault, so they can be addressed?

10/12/14 14:20:51 Exception in GZIP Filter: java.lang.NullPointerException
10/12/14 14:28:44 java.lang.OutOfMemoryError: Java heap space
...
10/12/14 14:28:55 java.lang.OutOfMemoryError: Java heap space
10/12/14 14:28:58 javax.servlet.ServletException: Java heap space


Thanks again!

Don,

We currently have an SR open on this issue (SR 3-2457305591) we do get the followin in the opmn.log though:

10/12/09 18:36:20 [libopmnoc4j] Could not connect to socket while pinging an OC4J process
10/12/09 18:36:20 [libopmnoc4j] Process Ping Failed: default_group~DEHYOASA~default_group~1 (346039309:788)
10/12/09 18:36:49 [pm-process] Process Crashed: default_group~DEHYOASA~default_group~1 (346039309:788) - Restarting

They had us change the ping timeout, no help, now they gave us this for the default_group:

<!-- Added by CNC -->
<category id="restart-parameters">
<data id="reverseping-timeout" value="600"/>
<data id="no-reverseping-failed-ping-limit" value="3"/>
<data id="reverseping-failed-ping-limit" value="6"/>
</category>

Put it in this weekend, so I don't know if it will work or not.

Tom

I believe GZIP is related to HTTP compression. If disabling compression is on option, you might try this as a "band-aid" until you can get a fix from Oracle.

Jeremy,

I think I'm beign dense today. Where do you see GZip or compression involved?

Tom

Hi,

Check that your antivirus is not scanning the OAS folder.

Tom, I see GZIP mentioned in the error message:

"Exception in GZIP Filter: java.lang.NullPointerException"

Sebastion,

Any particular OAS folder, or the entire ORACLE_HOME directory structure?

Tom

Hi,

I excluded the entire oc4j folder and subfolders.

I had a client with a "similar" problem. The issue was that the user starting the service was the one that I signed on with. When I would sign off of the server, periodically the services would stop.

We were told to create a user that is specific to starting the services only. Once I did that, there was no problems. This was on a windows server.

Maybe something to look at.