db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kathey Marsden <kmarsdende...@sbcglobal.net>
Subject Re: Working on adding a test case which will reliably run into sequence contention problem
Date Fri, 30 Sep 2011 22:36:52 GMT
On 9/30/2011 12:46 PM, Mamta Satoor wrote:
> Hi Kathey,
> Thanks for looking at the program. I forgot to include the deadlock
> timeout property in my email earlier that I used to test my program.
> Here is the command line I have tested with
> $ java -Dderby.locks.waitTimeout=3 -Dderby.locks.deadlockTimeout=2
> -Dderby.language.sequence.preallocator=5 DERBY4437Sequence 10 1000000
>> dellater.txt
> In fact, I even brought my trunk codeline to not include Rick's
> changes 1174290 so the sequence contention won't get wrapped around in
> a lock time out exception.But I still can't make the progrm run into
> sequence contention. I will keep working on it but if it time permit,
> I will appreicate you and others looking into the test.
I am sorry Mamta, I misunderstood your mail and thought you were working 
on  a small reproduction for the deadlock issue, DERBY-5430.  Instead I 
realize now you are working on a test case to cover DERBY-5423 / 
DERBY-5426.  I think it would be good to have  a testcase that covers 
the generation of this error.

I took a quick look at the program.  I didn't see a join() on the 
threads.  Perhaps the program is exiting prematurely as soon as all the 
threads get launched.

View raw message