Re: SKIP LOCKED on RAC

 I can't find the phrase in the manuals but the statement of RAC (and OPS before that) was that Global Cache locking was totally independent of (they may even have said "orthogonal to" row locking.

"really shouldn't be using skip locked" doesn't mean "isn't supported", it means "if you need to use skip locked you're probably going to be modifying the same data sets from multiple nodes and that's going to wreck performance" (And following on from Andy's comment about AQ - the reason why the guideline is to have instance-specific queues is that the critical structure is (was) and IOT, so every instance would be enqueueing to the same leaf block, and every instance would be dequeueing from the same leaf block if you had a global queue.

Regards
Jonathan Lewis

On Mon, 22 Nov 2021 at 22:17, Mladen Gogala <gogala.mladen_at_gmail.com> wrote:

> On 11/22/21 15:52, Andy Klock wrote:
> > Hi Mladen,
> >
> > "Mladen Gogala" <gogala.mladen_at_gmail.com> writes:
> >
> >> I know that SELECT FOR UPDATE SKIP LOCKED is not supported
> >> across the RAC instances but I am unable to find any Oracle
> >> documentation documenting that behavior. Can anybody help?
> > I can understand leaning towards writing the application to avoid
> > concurrent updates from different instances, but I wasn't aware
> > that SELECT FOR UPDATE SKIP LOCKED wasn't supported.
> >
> > Andy K
> >
> I was told that by an Oracle Support engineer after working on a SR for
> one of the customers. SKIP LOCKED was causing a nice little ORA-00600
> and the support guy told me that we really shouldn't be using SKIP
> LOCKED across the RAC instances. However, I cannot locate any support
> article stating that explicitly. ORA-00600 is an Oracle bug, by
> definition. Now, I would like to see that documented somewhere.
>
> --
> Mladen Gogala
> Database Consultant
> Tel: (347) 321-1217
> https://dbwhisperer.wordpress.com
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>

--
http://www.freelists.org/webpage/oracle-l