Обсуждение: Re: [COMMITTERS] pgsql: Document that a CHECKPOINT before taking a file system snapshot

2010/11/25 Bruce Momjian <bruce@momjian.us>:
> Document that a CHECKPOINT before taking a file system snapshot can
> reduce recovery time.

I didn't follow that on -hackers, but :

* checkpoint take place in the pg_start_backup process (before it
releases the hand, so before you can start snapshoting)

* we used to issue a checkpoint *before* pg_start_backup to reduce
pg_start_backup duration  in case you have wal_archiving turn off and
full_page_write is off, it reduces a bit the IO contention. (or even
if not I hate seing this pg_start_backup taking seconds/minutes to
finish)

How the checkpoint before will reduce recovery time ?


>
> Branch
> ------
> master
>
> Details
> -------
> http://git.postgresql.org/gitweb?p=postgresql.git;a=commitdiff;h=7276ab5888d85782d988fc297ad2e176c7ad1bca
>
> Modified Files
> --------------
> doc/src/sgml/backup.sgml |    2 ++
> 1 files changed, 2 insertions(+), 0 deletions(-)
>
>
> --
> Sent via pgsql-committers mailing list (pgsql-committers@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-committers
>



--
Cédric Villemain               2ndQuadrant
http://2ndQuadrant.fr/     PostgreSQL : Expertise, Formation et Support

C�dric Villemain wrote:
> 2010/11/25 Bruce Momjian <bruce@momjian.us>:
> > Document that a CHECKPOINT before taking a file system snapshot can
> > reduce recovery time.
>
> I didn't follow that on -hackers, but :
>
> * checkpoint take place in the pg_start_backup process (before it
> releases the hand, so before you can start snapshoting)
>
> * we used to issue a checkpoint *before* pg_start_backup to reduce
> pg_start_backup duration  in case you have wal_archiving turn off and
> full_page_write is off, it reduces a bit the IO contention. (or even
> if not I hate seing this pg_start_backup taking seconds/minutes to
> finish)
>
> How the checkpoint before will reduce recovery time ?

This is for using file system snapshots, not PITR or continuous
archiving.  The checkpoint reduces how much WAL has to be replayed.

--
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + It's impossible for everything to be true. +