Re: Locking mode in 10g seems to have changed

DA Morgan wrote:
> Comments in-line.
>
> claytonarends_at_gmail.com wrote:

> > Was this an Oracle oversight?
>
> No this is a bad design that, with changes in Oracle over the last
> decade has finally broken.
>
> > Is there a configuration setting to control this behavior?
>
> Wrong direction.
>
> > Any other thoughts on this?
>
> Get your team trained on 10g and rewrite your code to be scalable. If
> this was put on a RAC cluster it would beg for mercy.

If you follow Anurag's bug to the base bug (3646162) fix that caused it, this is an example of something that wasn't broke, but Oracle's implementation of RAC spotlighted a scalability defect in Oracle's locking mechanism, the fix of which broke a working bad design. So I'd say it was an Oracle oversight, compounded by application code dependent on a bug. Larry Ellison says the grid has been worked on for the last 13 years, so the oversight potentially dates from then.

So Clayton, now you know, now you can fix. We all get another example of how "doing things right" _does_ help avoid future problems.

jg

--
@home.com is bogus.
http://search.deviantart.com/searchcraft/?cmd=1&offset=24&search=corvette