Re: ALTER TABLE lock strength reduction patch is unsafe

Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
> On Mon, Jun 20, 2011 at 10:56 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> The ALTER TABLE patch
>> has greatly expanded the scope of the issue, and that *is* a regression
>> compared to prior releases.

> I agree the scope for RELOID errors increased with my 9.1 patch. I'm
> now happy with the locking patch (attached), which significantly
> reduces the scope - back to the original error scope, in my testing.

> I tried to solve both, but I think that's a step too far given the timing.

> It seems likely that there will be objections to this patch.

Yup, you're right. Having read this patch, I have absolutely zero
confidence in it. It introduces some locks in random places, with no
rhyme or reason that I can see. There is no reason to think that this
is a complete solution, and considerable reason to think that it isn't
(notably, the RELOID syscache is hardly the only one at risk). Worse,
it's adding more locking in performance-critical places, which seems
to me to severely degrade the argument for the original feature,
namely that it was supposed to give us *less* locking.

regards, tom lane