From: Peter Eisentraut <peter_e@gmx.net>
To: Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: Database system identifier from SQL
Date: Tue, 10 Jul 2007 20:06:10 +0200
User-Agent: KMail/1.9.5
Cc: pgsql-sql@postgresql.org
References: <200707101624.47885.peter_e@gmx.net>
	<20025.1184080999@sss.pgh.pa.us>
In-Reply-To: <20025.1184080999@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200707102006.11643.peter_e@gmx.net>

Tom Lane wrote:
> Peter Eisentraut <peter_e@gmx.net> writes:
> > Is there a way to query the database system identifier that
> > pg_controldata outputs from SQL?
>
> Don't think so.  Do you have a use-case for providing a function to
> return that?

I'd like to find out whether two connections are really to the same 
server.

This is in the context of an installation routine for a database 
application which pulls the connection parameters for several databases 
from a configuration file and creates various things there.  It should, 
however, only create roles once if the connection parameters point to 
the same server (if the user only expects a small installation, say).

A workaround is to compare inet_server_addr() and inet_server_port(), 
but that is not 100% safe because it doesn't cover Unix-domain sockets 
using different paths.

-- 
Peter Eisentraut
http://developer.postgresql.org/~petere/