From: | Cédric Villemain <cedric(dot)villemain(at)dalibo(dot)com> |
---|---|
To: | pgsql-admin(at)postgresql(dot)org |
Cc: | "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>, Chris Hoover <revoohc(at)gmail(dot)com>, Andrzej Zawadzki <zawadaa(at)gmail(dot)com> |
Subject: | Re: Moving database install to new SAN |
Date: | 2007-09-21 12:29:42 |
Message-ID: | 200709211429.47837.cedric.villemain@dalibo.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
Le jeudi 20 septembre 2007, Joshua D. Drake a écrit :
> Chris Hoover wrote:
> >>> 2. copy whole data directory over to new SAN
> >>
> >> If database is big, then I think that is much faster to copy running
> >> database
> >> to new SAN.
> >> Then stop postgres and rsync what was changed - but this will be
> >> probably only a few files.
> >
> > I believe if you do this, you will not get a function database in the
> > end. There is a lot of data that is held in memory/buffers that may not
> > be flushed to the disks. You have no guarantee you will get this data
> > with this method.
>
> The method would work if they are willing to have an outage. Basically
> you do an initial rsync of the large db...
Perhaps using some tablespace to move the bigest data, then rsync ...
>
> Then you shut down the db.
>
> And rsync again, which will be faster than doing a complete move with
> shutdown.
>
> The key here is, if you use this method... there is *zero* way around
> shutting down the database before the second rsync.
>
>
> Sincerely,
>
> Joshua D. Drake
--
Cédric Villemain
Administrateur de Base de Données
Cel: +33 (0)6 74 15 56 53
http://dalibo.com - http://dalibo.org
From | Date | Subject | |
---|---|---|---|
Next Message | Alvaro Herrera | 2007-09-21 13:22:41 | Re: How to repair if file pg_clog/0000 is deleted |
Previous Message | Yogesh Arora | 2007-09-21 05:53:43 | Schedular Problem in Postgres |