Robert Haas wrote:
> On Fri, Jun 17, 2016 at 9:33 AM, Andrew Gierth
> <> wrote:
> >>>>>> "Robert" == Robert Haas <> writes:
> >  >> Why is the correct rule not "check for and ignore pre-upgrade mxids
> >  >> before even trying to fetch members"?
> >
> >  Robert> I entirely believe that's the correct rule, but doesn't
> >  Robert> implementing it require a crystal balll?
> >
> > Why would it? Pre-9.3 mxids are identified by the combination of flag
> > bits in the infomask, see Alvaro's patch.
> I see the patch, but I don't see much explanation of why the patch is
> correct, which I think is pretty scary in view of the number of
> mistakes we've already made in this area.

... and actually the patch fails one isolation tests in 9.3, which is
why I haven't pushed (I haven't tested 9.4 but I suppose it should be
the same).  I'm looking into that now.

> The comments just say:
> + * A tuple that has HEAP_XMAX_IS_MULTI and HEAP_XMAX_LOCK_ONLY but neither of
> + * XMAX_EXCL_LOCK and XMAX_KEYSHR_LOCK must come from a tuple that was
> + * share-locked in 9.2 or earlier and then pg_upgrade'd.
> Why must that be true?

The reason this must be true is that in 9.2 and earlier multixacts were only
used to lock tuples FOR SHARE, which had that specific bit pattern.  I suppose
I need to make this comment more explicit.

> + * We must not try to resolve such multixacts locally, because the result 
> would
> + * be bogus, regardless of where they stand with respect to the current valid
> + * range.
> What about other pre-upgrade mxacts that don't have this exact bit
> pattern?  Why can't we try to resolve them and end up in trouble just
> as easily?

There shouldn't be any.  Back then, it was not possible to have tuples
locked and updated at the same time; FOR UPDATE (the only other locking
mode back then) didn't allow other lockers, so the only possibility was
FOR SHARE with that bit pattern.

Álvaro Herrera      
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Sent via pgsql-hackers mailing list (
To make changes to your subscription:

Reply via email to