Re: Streaming replication: rsync to switchover

From: Gerhard Hintermayer <gerhard(dot)hintermayer(at)gmail(dot)com>
To: rd <rudi(dot)pettazzi(at)kemen(dot)it>, pgsql-admin <pgsql-admin(at)postgresql(dot)org>
Subject: Re: Streaming replication: rsync to switchover
Date: 2011-04-19 08:09:04
Message-ID: BANLkTikrqROJjNiVpmkZorkPM7nUYHZUAw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

Hi, I do rsync -a --delete new_primary_server::postgresql-data/
/var/lib/postgresql/9.0/data/

Take care to use -a, I had -r and wondered, why the rsync took so long,
despite the data was nearly the same.

assuming you have configured rsyncd on the server as:

[postgresql-data]
uid = postgres
path = /var/lib/postgresql/9.0/data
comment = PostgreSQL 9.0 data dir
exclude = postmaster.log pg_xlog/* postmaster.pid postgresql.conf

The exclude might be tunable a litte (the confs eg. are not stored in the
datadir anymore on my distro)
Always rsync over the existing data dir (of course stop server before)
The former master can also be stopped later. When the trigger file is
created, the connection to it is closed anyhow.
Don't forget to pg_start_backup pg_stop_backup on the new server
before/after you do rsync.

Gerhard

On Mon, Apr 18, 2011 at 7:16 PM, rd <rudi(dot)pettazzi(at)kemen(dot)it> wrote:

> Hi,
>
> I've setup streaming replication + file-based log-shipping, with a
> hot-standby.
>
> Everything's good, except I don't know which is the correct way to sync the
> once-master db with the newly promoted master after a switchover,
>
> 1) I shutdown the master and touch the trigger_file on the slave. Now the
> slave becomes the master.
> 2) On the old-master I create recovery.conf + archive directory for
> file-based log-shipping.
> 3) I hope I can setup the old-master with a simple rsync, because the
> operation could be really fast, being M and S almost identical, but I'm not
> sure about the correct way to rsync them:
>
> a) should I rsync --delete?
> b) should I delete the pg_xlog of the old master, before rsyncing? (I'm
> wondering what could happen if some of them never made its way to the slave,
> that now has become new-master)
> c) am I missing something? For example, I've read somewhere that one should
> copy pg_control and pg_stat from the old slave to the new slave but it
> doesn't make sense to me.
>
>
> thanks for any hint,
>
>
> --
> rudi pettazzi
>
> --
> Sent via pgsql-admin mailing list (pgsql-admin(at)postgresql(dot)org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-admin
>

In response to

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Selva manickaraja 2011-04-19 08:40:42 Orphan Temp Table After PITR
Previous Message Muruganand Karthikeyan 2011-04-19 07:08:37 Re: Where is the source code for v8.4.5