Default Server Restarting...Help needed

[Neill Rynders]

Environment : Windows 2000 SP2
WebSphere 4.3 with IIS and WAS_Plugin_10-18-2002_4.0.x_cumulative
3 Enterprise applications are running in the Default server. In the tracefile we see this message: [17/11/02 13:15:18:975 AST] 16df70ce ActiveServerP A ADMS0028I: Restarting server: Default Server
[17/11/02 13:15:19:100 AST] 16df70ce ActiveServerP A ADMS0032I: Started server: Default Server (pid 2368)
There are no errors in the STD_error logs to indicate failure reason. No errors are seen on the console at the time of the restart either.
1) Is there any IBM document to indicate under what conditions a server will restart itself ?
2) Does anyone know how to troubleshoot / trace the reason for the restart ?
There is no fixed pattern to the failures as they occur under different conditions and are not easily recreatable.
Thanks in advance.

[Patrick Finnegan]

If the JVM crashes the app server will try to restart it.
Check the windows event log and any Dr Watsons. Also check the tracefile and activity log. Chapter 24 of the Websphere handbook covers troble shooting.
Downloadable in PDF fomat from Available from:
http://publib-b.boulder.ibm.com/Redbooks.nsf/RedbookAbstracts/sg246176.html?Open

[Neill Rynders]

Thanks for the info, found it very usefull.
We have finally nailed down the problem to this :
Enabled Verbose Garbage Collection.
Just before the restart of the Default server, this message shows up in the log:
<GC[0]: Expanded System Heap by 65536 bytes>
<GC[0]: Expanded System Heap by 65536 bytes>
<GC[0]: Expanded System Heap by 65536 bytes>
And after this we have the restart of the default server. So, is seems there is an attempt to increase the Native stack memory ??? and not the application stack from what I can see. There are several messages indicating an Allocation Failure, then a normal garbage collection run is triggered which ends successfully.
Has anyone seen this error before ? There seems to be very little information available on tracing anything in the Native Stack of the JVM.
Thanks.

[Neill Rynders]

Sorry, the exact error message seems to have been lost in the post, here is the error message:
GC[0]: Expanded System Heap by 65536 bytes
GC[0]: Expanded System Heap by 65536 bytes
GC[0]: Expanded System Heap by 65536 bytes

[Patrick Finnegan]

First load all efixes available for 4.0.3.
They can be downloaded from:
http://www-1.ibm.com/support/search.wss?rs=180&tc=SSEQTP&dc=D400
Search all versions using the keyword "4.0.3".
Regards.
Patrick.

[Shalini Komarla]

Seems like the jvm is running out of heap meomory. If your system config allows, you could try increasing your heap size using the -mx & -ms parameters in websphere.

[Khurram Fakhar]

Hi Friends,
I tried to install WSphere 4.0.1 on windows 2000 .
It took my Network userID and password for the first time , and when second time i tried to re-install the server then it was not accepting the Network user id and password , so i entered local Administrator login of the system ..

[Neill Rynders]

Well, after a lot of troubleshooting we still have no official answer to this restart. We did however achieve a more stable environment falling back to a configuration with DB2 7.1 fixpack 3, and WebSphere 4.0.1 with no fixpacks. It seems there is something in the fixpacks which does not quite agree with our applications.

We do not have the restart / Java process crash for any of the normal reasons like JVM memory running out etc. There is not even a serious event thrown before the restart, no error log entries, nothing. Nice 1 to try and troubleshoot...not !!! To all those who replied, my sincere thanks.