Re: Fixing pg_basebackup with tablespaces found in $PGDATA - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: Fixing pg_basebackup with tablespaces found in $PGDATA
Date
Msg-id 20140110162401.GE4873@momjian.us
Whole thread Raw
In response to Re: Fixing pg_basebackup with tablespaces found in $PGDATA  (Bernd Helmle <mailings@oopsware.de>)
List pgsql-hackers
On Thu, Jan  2, 2014 at 03:34:04PM +0100, Bernd Helmle wrote:
> 
> 
> --On 1. Januar 2014 23:53:46 +0100 Dimitri Fontaine
> <dimitri@2ndQuadrant.fr> wrote:
> 
> >Hi,
> >
> >As much as I've seen people frown upon $subject, it still happens in the
> >wild, and Magnus seems to agree that the current failure mode of our
> >pg_basebackup tool when confronted to the situation is a bug.
> >
> >So here's a fix, attached.
> 
> I've seen having tablespaces under PGDATA as a policy within several
> data centres in the past. The main reasoning behind this seems that
> they strictly separate platform and database administration and for
> database inventory reasons. They are regularly surprised if you tell
> them to not use tablespaces in such a way, since they absorbed this
> practice over the years from other database systems. So +1 for
> fixing this.

FYI, this setup also causes problems for pg_upgrade.  There is a recent
thread about that that I will reply to.  The problem is that pre-9.2
servers get a mismatch between the symlink and the pg_tablespace path
when they rename the old cluster.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + Everyone has their own god. +



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [ADMIN] Column missing from pg_statistics
Next
From: Florian Pflug
Date:
Subject: Re: [PATCH] Negative Transition Aggregate Functions (WIP)