Re: PostgreSQL 9.6 wals management

From: Mariel Cherkassky <mariel(dot)cherkassky(at)gmail(dot)com>
To: Michael Paquier <michael(dot)paquier(at)gmail(dot)com>
Cc: PostgreSQL mailing lists <pgsql-performance(at)postgresql(dot)org>
Subject: Re: PostgreSQL 9.6 wals management
Date: 2017-11-20 12:56:27
Message-ID: CA+t6e1=w_R=vc2e2JZR_20YgPXJWTM6KEZmBDovvnSy+-NM6tg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Thank you for the clarification.

בתאריך 20 בנוב׳ 2017 14:28,‏ "Michael Paquier" <michael(dot)paquier(at)gmail(dot)com>
כתב:

> On Mon, Nov 20, 2017 at 6:02 PM, Mariel Cherkassky
> <mariel(dot)cherkassky(at)gmail(dot)com> wrote:
> > This morning , I set the wal_keep_segments to 100 and I set the
> > archive_timeout to 6 minutes. Now, after setting those settings and
> starting
> > the cluster wals switch is working fine and I didnt see that many wals
> were
> > However, doesnt the old wals should be deleted automaticly ? Can I
> > delete archives safely ?
>
> Archives are useful if they can be used with a base backup which would
> allow it to recover close to the point has created WAL activity, up to
> the last finished segment to be precise. So if you have no base
> backups or standbys (for example disconnected for a long) that would
> use them, there is no point in keeping them. What defines the archive
> and base backup retention is your data retention policy. Do not touch
> the files of pg_xlog though, those are managed by PostgreSQL itself.
> It is also good practice to put the archives on a different partition,
> and to not have the archives in a sub-path of the main data folder as
> you do as those would get included in all base backups taken.
> --
> Michael
>

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Stephen Frost 2017-11-20 14:33:05 Migration to pglister - Before
Previous Message Michael Paquier 2017-11-20 12:28:52 Re: PostgreSQL 9.6 wals management