Hi,

From: Simon Riggs <[EMAIL PROTECTED]>
Subject: Re: [Pgpool-general] Online recovery by PITR questions
Date: Wed, 14 May 2008 07:53:49 +0100

> On Thu, 2008-05-08 at 18:39 +0200, Nico -telmich- Schottelius wrote:
> 
> > pgpool executes CHECKPOINT on a master node
> 
> This step is not required, since a checkpoint is forced during
> pg_start_backup().

Yes, CHECKPOINT is executed in pg_start_backup(). But we do by two
reasons.

1) If PostgreSQL is 7.4, we cannot use PITR. Pgpool do CHECKPOINT to
flush buffer pool. Then pgpool copies database cluster.

  1) CHECKPOINT
  2) rsync
  3) stop all connections
  4) CHECKPOINT
  5) rsync

2) In PostgreSQL 8.3, pg_start_backup() takes a long time without
CHECKPOINT.

24.3.2. Making a Base Backup
http://www.postgresql.org/docs/8.3/static/continuous-archiving.html

So we execute CHECKPOINT explicitly.

Regards,
--
Yoshiyuki Asaba
[EMAIL PROTECTED]
_______________________________________________
Pgpool-general mailing list
[email protected]
http://pgfoundry.org/mailman/listinfo/pgpool-general

Reply via email to