Re: Reducing Catalog Locking

On 2014-10-31 09:48:52 -0400, Tom Lane wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > On a related note, I've previously had the thought that it would be
> > nice to have a "big DDL lock" - that is, a lock that prevents
> > concurrent DDL without preventing anything else - so that pg_dump
> > could get just that one lock and then not worry about the state of the
> > world changing under it.
>
> Hm ... how would that work exactly? Every DDL operation has to take
> the BigDDLLock in shared mode, and then pg_dump takes it in exclusive
> mode?
>
> That would preclude two pg_dumps running in parallel, which
> maybe isn't a mainstream usage but still there's never been such a
> restriction before. Parallel pg_dump might have an issue in particular.

It should probably be a heavyweight lock. Then every DDL operation can
take it in RowExclusiveLock mode and pg_dump can take ShareLock. As
RowExclusive is a fastpath elegible lock, that'll not even hit the
global lock table most of the time.

> But more to the point, this seems like optimizing pg_dump startup by
> adding overhead everywhere else, which doesn't really sound like a
> great tradeoff to me.

Well, it'd finally make pg_dump "correct" under concurrent DDL. That's
quite a worthwile thing.

Greetings,

Andres Freund

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