Re: pgsql: Fix a couple of bugs in MultiXactId freezing

On 2013-12-03 12:22:33 -0500, Tom Lane wrote:
> Andres Freund <andres(at)2ndquadrant(dot)com> writes:
> >> On 2013-12-03 09:48:23 -0500, Tom Lane wrote:
> >>> Is this bad enough that we need to re-wrap the release?
>
> > After looking, I think I am revising my opinion. The broken locking
> > behaviour (outside of freeze, which I don't see how we can fix in time),
> > is actually bad.
> > Would that stop us from making the release date with packages?
>
> That's hardly answerable when you haven't specified how long you
> think it'd take to fix.

There's two things that are broken as-is:
1) the freezing of multixacts: The new state is much better than the old
one because the old one corrupted data, while the new one somes removes
locks when you explicitly FREEZE.
2) The broken locking behaviour in Noah's test without the
FREEZE.

I don't see a realistic chance of fixing 1) in 9.3. Not even sure if it
can be done without changing the freeze WAL format. But 2) should be fixed
and basically is a oneliner + comments + test. Alvaro?

> In general, though, I'm going to be exceedingly unhappy if this release
> introduces new regressions. If we have to put off the release to fix
> something, maybe we'd better do so. And we'd damn well better get it
> right this time.

I think that's really hard for the multixacts stuff. There's lots of
stuff not really okay in there, but we can't do much about that now :(

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services