From: Tom Lane <tgl@sss.pgh.pa.us>
To: Holger Jakobs <holger@jakobs.com>
cc: "[PGSQL-ADMIN-LIST]" <pgsql-admin@postgresql.org>
Subject: Re: Export and import of statistics data
In-reply-to: <b97f9345-e617-e555-a6e9-7b185f2b17e2@jakobs.com>
References: <b97f9345-e617-e555-a6e9-7b185f2b17e2@jakobs.com>
Comments: In-reply-to Holger Jakobs <holger@jakobs.com>
	message dated "Tue, 23 Feb 2021 19:53:17 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <3891952.1614108750.1@sss.pgh.pa.us>
Date: Tue, 23 Feb 2021 14:32:30 -0500
Message-ID: <3891953.1614108750@sss.pgh.pa.us>
Precedence: bulk

Holger Jakobs <holger@jakobs.com> writes:
> Is there a way of exporting statistics data from a database cluster 
> containing a lot of rows and import them into a similar structured, but 
> small database cluster.

Not at present, AFAIK.  There have been discussions of creating a
way to do that, and I know of at least one non-public tool that
does it, but nobody's pushed things to the point of having code
the community would accept.

The key stumbling block is inventing a dump representation that
would have realistic prospects of still being importable if/when
we change the contents of pg_statistic.  Certainly there are a
lot of applications where cross-version compatibility isn't a
requirement; but I don't think the community would take code that
isn't usable by pg_upgrade, and for that you need an upgradable
design.

			regards, tom lane