From: Tom Lane <tgl@sss.pgh.pa.us>
To: Ayush Vatsa <ayushvatsa1810@gmail.com>
cc: pgsql-general@lists.postgresql.org
Subject: Re: Clarification on Role Access Rights to Table Indexes
In-reply-to: <CACX+KaMz2ZoOojh0nQ6QNBYx8Ak1Dkoko=D4FSb80BYW+o8CHQ@mail.gmail.com>
References: <CACX+KaMz2ZoOojh0nQ6QNBYx8Ak1Dkoko=D4FSb80BYW+o8CHQ@mail.gmail.com>
Comments: In-reply-to Ayush Vatsa <ayushvatsa1810@gmail.com>
	message dated "Mon, 17 Feb 2025 23:31:46 +0530"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <855987.1739816850.1@sss.pgh.pa.us>
Date: Mon, 17 Feb 2025 13:27:30 -0500
Message-ID: <855988.1739816850@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/855988.1739816850%40sss.pgh.pa.us>
Precedence: bulk

Ayush Vatsa <ayushvatsa1810@gmail.com> writes:
> postgres=> SELECT pg_prewarm('pg_class_oid_index');
> ERROR:  permission denied for index pg_class_oid_index

You'd really have to take that up with the author of pg_prewarm.
It's not apparent to me why checking SQL access permissions is
the right mechanism for limiting use of pg_prewarm.  It seems
like ownership of the table would be more appropriate, or maybe
access to one of the built-in roles like pg_maintain.

> 1. Can a role have access rights to a table without having access to its
> index?

Indexes do not have access rights of their own, which is why
access rights are a poor gating mechanism for something that
needs to be applicable to indexes.  Ownership could work,
because we make indexes inherit their table's ownership.

			regards, tom lane