Fwd: Sharing data directories between machines

---------- Forwarded message ----------
From: John McKown <john.archie.mckown@gmail.com>
Date: Tue, Aug 6, 2013 at 2:23 PM
Subject: Re: [GENERAL] Sharing data directories between machines
To: Davide Setti <davide.setti@gmail.com>


Given that the user seems to want a single copy of the data base, rather than two replicated and synchronized copies, wouldn't that indicate that the clients on all machines need to point to the PostgreSQL server on one machine? Instead of trying to have a the data base server process running on both machines and pointing at the same physical data base. The thought makes me shudder.

If it were me (on Linux or UNIX, I don't know Windows), I'd have a world-readable file somewhere on the "mounted" drive (shared via NFS or CIFS I guess). This file would simply contain lines like:

export PGHOST=ip.or.name.of.machine.running.server
export PGPORT=5432 #stand PostgreSQL TCP/IP port

I would then "source" that file in the shell's start up script on all machines, and also "source" it in the start up scripts for everything else which uses PostgreSQL. This shared file could be modified whenever the PostgreSQL server is moved from one machine to another.

Too bad the PostgreSQL server cannot be in a "Federated" configuration so that it "knows" which databases are controlled by which server and automatically passes the requests around. Or does it and I just can't find it?
http://en.wikipedia.org/wiki/Federated_database_system
IBM's DB2 can do this.
--

As of next week, passwords will be entered in Morse code.

Maranatha! <><
John McKown