Feedback-ID: ia2694551:Fastmail
Date: Thu, 25 Apr 2024 10:25:49 +0200
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Peter Eisentraut <peter@eisentraut.org>
Cc: Nicolas Seinlet <nicolas@seinlet.eu>,
	"pgsql-docs@lists.postgresql.org" <pgsql-docs@lists.postgresql.org>
Subject: Re: wal_init_zero / wal_recycle
Message-ID: <202404250825.uotokqhaopxu@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <6a09dfdd-9dc5-49f4-82ea-826162ed0d74@eisentraut.org>
Archived-At: 
 <https://www.postgresql.org/message-id/202404250825.uotokqhaopxu%40alvherre.pgsql>
Precedence: bulk

On 2024-Apr-24, Peter Eisentraut wrote:

> On 24.04.24 10:08, Nicolas Seinlet wrote:

> > It was not crystal clear to me that when wal_recycle is also ON , the
> > recycled files are not filled with zeroes, because they are not
> > considered as new. Could we add a sentence stating this?
> 
> It currently says:
> 
> wal_init_zero: "... causes new WAL files to be filled with zeroes ..."
> 
> wal_recycle: "... avoiding the need to create new ones ..."
> 
> I'm not sure where exactly to make that more clear.

It may be too subtle.  Maybe we could add a note at the end:

wal_init_zero (boolean)

If set to on (the default), this option causes new WAL files to be filled with
zeroes. On some file systems, this ensures that space is allocated before we
need to write WAL records. However, Copy-On-Write (COW) file systems may not
benefit from this technique, so the option is given to skip the unnecessary
work. If set to off, only the final byte is written when the file is created so
that it has the expected size.  >>Note that recycled WAL files are for these
purposes not considered new.<<

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"Pero la cosa no es muy grave ..."     (le petit Nicolas -- René Goscinny)