Re: how to determine cause of deadlock

Gdas,

Glad to know you solved your problem.

Steve Adams' site has quite a few Q&A's discussing deadlocks. I read all of them on there because last Saturday we had 4 deadlocks in udump file (we used to 1 every two weeks on average). It seems increasing INITTRANS is a solution to *some* cases. I guess a deadlock caused by two transactions locking each other (as shown in your udump file snippet posted on 07/13) can be solved by increasing INITTRANS. Another kind of lock, self-deadlock, is probably more difficult. I have that problem once in a while. For debugging, I need to set events 4020 on it, according to Steve Adams' suggestion.

Yong Huang

gdas <gdas1NOgdSPAM_at_yahoo.com.invalid> wrote in message news:15d5d80c.b9c2a2e6_at_usw-ex0102-015.remarq.com...
> Jonathan,
>
> I rebuilt the table and the indexes with a higher initrans value
> (I put both at 100...no scientific method here...just randomly
> picked it).
>
> The application has now been running for 3 consecutive days and
> not a single deadlock has occurred. Now that we've proven the
> solution, I'll go back and try to set a lower initrans value to
> minimize the wasted space.
>
> Who would've thought? Learning something new everyday...
>
> Thanks for your help,
> Gavin
>
>
>
> -----------------------------------------------------------
>
> Got questions? Get answers over the phone at Keen.com.
> Up to 100 minutes free!
> http://www.keen.com
>
Received on Tue Jul 18 2000 - 00:00:00 CDT