Re: MultiXact truncation, startup et al.

On 2013-11-28 19:23:29 -0500, Robert Haas wrote:
> On Wed, Nov 27, 2013 at 5:42 PM, Andres Freund <andres@2ndquadrant.com> wrote:
> > So, I've done this for 9.3+ for now. Testing around that turned up that
> > our current way to schedule anti mxid wraparounds doesn't really work:
> > 1) autovacuum.c knows about such vacuums, but vacuum.c doesn't. Leading
> > to a long cycle of partial vacuums that don't increase relminmxid.
> > 2) Parts of the code used 200mio as a hardcoded constant, others used
> > autovacuum_freeze_max_age.
> >
> > 0001 fixes the vacuum scheduling and is applicable to 9.3+,
> 
> multiTableLimit is a bad name for whatever concept this is supposed to
> represent.  It does not involve multiple tables.

It's the freezeTableLimit equivalent for multixacts. I haven't named it
multiFreezeTableLimit because multixacts don't actually have a concept
of freezing - and in that context I didn't see much danger for it to be
understood to apply to multiple tables.

I'll try to think of a better name.

> vacuum_set_xid_limits now has multiXactCutoff, multiTableLimit, and
> mxLimit, and there's no explanation of what the are.

Well, neither have the plain xid variants. Not that that's good, but
it's not this patches fault.
I'd be happy to provide a separate patch that adds documentation for
each of them - I've wondered often enough about their meaning to make it
rather worthwhile.

> > 0002 re-adds pg_multixact truncation during crash recovery. The current
> > code will only work on 9.3+, but if it's deemed acceptable I can
> > backport it to earlier versions. I am not sure if it's worth backporting
> > it 9.0 given it has neither HS nor SR?
> 
> Huh?

That should have been a < 9.0 aka. 8.4.

Patch 02 has changed its shape slightly since the version I posted here,
because it's a prerequisite for the fix for the multixact bugs around
heap_freeze_tuple() and heap_tuple_needs_freeze() I've written about
nearby. I think Alvaro plans to work over my fixes to make them look
nice enough and commit them soonish.
Should somebody want to look into it
http://git.postgresql.org/gitweb/?p=users/andresfreund/postgres.git;a=shortlog;h=refs/heads/multixact-handling
contains the fixes.

Greetings,

Andres Freund

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