Date: Mon, 24 Apr 2023 19:18:54 -0500
From: Justin Pryzby <pryzby@telsasoft.com>
To: Michael Paquier <michael@paquier.xyz>
Cc: Soumyadeep Chakraborty <soumyadeep2007@gmail.com>,
	Zhihong Yu <zyu@yugabyte.com>, pgsql-hackers@postgresql.org,
	Ashwin Agrawal <ashwinstar@gmail.com>, vanjared@vmware.com
Subject: Re: ALTER TABLE SET ACCESS METHOD on partitioned tables
Message-ID: <ZEccbkBYCMy42Rxt@telsasoft.com>
References: <ZCIxFkgOGv64nYE7@paquier.xyz>
 <ZCJuW3GOfE/0eBjh@telsasoft.com>
 <ZCUZLiCeXq4Im7OG@telsasoft.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZCUZLiCeXq4Im7OG@telsasoft.com>
Archived-At: <https://www.postgresql.org/message-id/ZEccbkBYCMy42Rxt%40telsasoft.com>
Precedence: bulk

On Thu, Mar 30, 2023 at 12:07:58AM -0500, Justin Pryzby wrote:
> On Mon, Mar 27, 2023 at 11:34:36PM -0500, Justin Pryzby wrote:
> > On Tue, Mar 28, 2023 at 09:13:10AM +0900, Michael Paquier wrote:
> > > On Mon, Mar 20, 2023 at 09:30:50AM +0900, Michael Paquier wrote:
> > > > Did you check dump and restore flows with partition
> > > > trees and --no-table-access-method?  Perhaps there should be
> > > > some regression tests with partitioned tables?
> > > 
> > > I was looking at the patch, and as I suspected the dumps generated
> > > are forgetting to apply the AM to the partitioned tables.
> > 
> > The patch said:
> > 
> > +       else if (RELKIND_HAS_TABLE_AM(relkind) || relkind == RELKIND_PARTITIONED_TABLE)
> > 
> > pg_dump was missing a similar change that's conditional on
> > RELKIND_HAS_TABLE_AM().  It looks like those are the only two places
> > that need be be specially handled.
> > 
> > I dug up my latest patch from 2021 and incorporated the changes from the
> > 0001 patch here, and added a test case.
> > 
> > I realized that one difference with tablespaces is that, as written,
> > partitioned tables will *always* have an AM specified,  and partitions
> > will never use default_table_access_method.  Is that what's intended ?
> > 
> > Or do we need logic similar tablespaces, such that the relam of a
> > partitioned table is set only if it differs from default_table_am ?
> 
> Actually .. I think it'd be a mistake if the relam needed to be
> interpretted differently for partitioned tables than other relkinds.
> 
> I updated the patch to allow intermediate partitioned tables to inherit
> relam from their parent.
> 
> Michael wrote:
> > .. and there are quite more points to consider.
> 
> What more points ?  There's nothing that's been raised here.  In fact,
> your message last week is the first comment since last June ..

Michael ?