Re: [Wrapper-user] SERVICE_CONTROL_INTERROGATE - last message before wrapper dies

[Krishna K. <kko...@wo...>]

Just wanted to update the group regarding this problem.  It turned out 
that we were specifying memory arguments to the JVM in the wrapper's 
config. file erroneously.

The JVM argument parameters were provided twice:

wrapper.java.additional.4=-Xms256m
wrapper.java.additional.5=-Xmx800M


# Initial Java Heap Size (in MB)
wrapper.java.initmemory=25

# Maximum Java Heap Size (in MB)
wrapper.java.maxmemory=100

The result is that the command line that wrapper process generated 
looked like this:

java -Xms256m -Xmx800M -Xms25m -Xmx100m

The second set of memory arguments were not enough for our process to 
run and the service would abend due to insufficient memory and the 
OutOfMemory error line would also not show up in the logs.


Leif Mortenson wrote:

> Krishna Kottamasu wrote:
>
>> Thanks for your prompt response.   Upgrading this customer to 
>> 3.1.1/3.1.2 could take some time.  I can try setting 
>> wrapper.debug=false in the config. file and see how it goes.  From 
>> your message, this is what I understood:
>
>
> Ok, I assumed you had enabled debug output when you started 
> encountering problems.
> Is there a reason why you have debug output enabled by default?  Debug 
> output builds
> up pretty large log files in a hurry.  It looks like you have it 
> limited to 10, 10MB files
> though so you should be Ok as far as disk-space.
>
>> If more than one thread in our application tries to write something 
>> to the standard out (which the wrapper traps and outputs to the 
>> wrapper's log file) at the same time, that could be crashing the 
>> wrapper process.  Is my understanding correct?
>
>
> I want to reclarify this.   It is only a problem with multithreading 
> in the C code of the
> Wrapper.exe process.   There is NOT a problem when console log output 
> coming
> from the Java program itself.   Logging from multi threaded Java apps 
> is perfectly
> safe.
>
> If the User presss CTRL-C,  second thread from the OS is called into 
> the Wrapper
> process.   If my debug log output from this second thread attempts to 
> log at the same
> instant as a log message from the Wrapper's main event loop then there 
> was a chance
> of memory corruption.   With debug output disabled, this would pretty 
> much never
> happen.
>
>> I've also answered your questions right underneath them in the 
>> original message.
>
>
> Thanks,
> Leif
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM. 
> Deadline: Sept. 13. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Wrapper-user mailing list
> Wra...@li...
> https://lists.sourceforge.net/lists/listinfo/wrapper-user
>
>