Horton

Re: pg_dump is broken for partition tablespaces

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: David Rowley <david.rowley@2ndquadrant.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Alvaro Herrera <alvherre@2ndquadrant.com>, Andres Freund <andres@anarazel.de>
Date: 2019-03-06T07:19:00Z
Lists: pgsql-hackers 
On Wed, Mar 06, 2019 at 07:45:06PM +1300, David Rowley wrote:
> Partitioned indexes have this similar inherit tablespace from parent
> feature, so ca4103025dfe26 was intended to align the behaviour of the
> two. Partitioned indexes happen not to suffer from the same issue as
> the indexes are attached after their creation similar to what I
> propose above.
> 
> Can anyone see any fundamental reason that we should not create a
> partitioned table by doing CREATE TABLE followed by ATTACH PARTITION?
>  If not, I'll write a patch that fixes it that way.

The part for partitioned indexes is already battle-proven, so if the
part for partitioned tables can be consolidated the same way that
would be really nice.

> As far as I can see, the biggest fundamental difference with doing
> things this way will be that the column order of partitions will be
> preserved, where before it would inherit the order of the partitioned
> table.  I'm a little unsure if doing this column reordering was an
> intended side-effect or not.

I don't see any direct issues with that to be honest thinking about
it..
--
Michael

Commits

  1. Make pg_dump emit ATTACH PARTITION instead of PARTITION OF (reprise)

  2. Fix tablespace inheritance for partitioned rels

  3. Make pg_dump emit ATTACH PARTITION instead of PARTITION OF

  4. Make sure ALTER TABLE preserves index tablespaces.