Re: prevent WAL replication to fill filesystem

From: Ninad Shah <nshah(dot)postgres(at)gmail(dot)com>
To: Jehan-Guillaume de Rorthais <jgdr(at)dalibo(dot)com>
Cc: Laurenz Albe <laurenz(dot)albe(at)cybertec(dot)at>, basti <mailinglist(at)unix-solution(dot)de>, pgsql-general <pgsql-general(at)postgresql(dot)org>
Subject: Re: prevent WAL replication to fill filesystem
Date: 2021-09-08 04:07:15
Message-ID: CAOFEiBdsFRoSuJs+A3cJ3EFFV4QLLJEHcr_w8GSrC2h8Sxo4Zw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

These suggestions are appropriate. However, if you are ready to lose your
replica to keep production running, there are a couple of methods.

1) Set archive_command to "/bin/true"
2) rename .ready files in archive_status to .done.

Again, I personally do not recommend this, so consider this as a last
resort to save your system from the space crunch issue.

Regards,
Ninad Shah

On Tue, 31 Aug 2021 at 15:08, Jehan-Guillaume de Rorthais <jgdr(at)dalibo(dot)com>
wrote:

> On Tue, 31 Aug 2021 10:53:45 +0200
> Laurenz Albe <laurenz(dot)albe(at)cybertec(dot)at> wrote:
>
> > On Tue, 2021-08-31 at 10:36 +0200, basti wrote:
> > > have a old PG 9.6 with WAL replication.
> > > for some reason the rsync was failed and the filesystem was filled by
> > > pg_xlog files.
> > > As result PG stops working.
> > >
> > > Is there a way to prevent WAL replication to fill filesystem by xlog
> > > files in case of failure?
>
> No.
>
> You have to size your partition big enough to handle few days of normal
> WAL production, monitor the archiving success and react quickly to fix
> things
> when it fails.
>
> > Upgrade to v13 and set "max_slot_wal_keep_size".
>
> How this would prevent the failure if $OP doesn't (want to) use slots ? He
> speaks about rsync, so it might be purely archiving error, not streaming.
>
> Regards,
>
>
>

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message hubert depesz lubaczewski 2021-09-08 04:59:19 Is there any technical reason why "alter table .. set not null" can't use index?
Previous Message Michael Lewis 2021-09-07 15:27:07 Re: Query takes around 15 to 20 min over 20Lakh rows