Message-ID: <8177cb9bc419f96f09302c50eb71d92723e01b9e.camel@cybertec.at>
Subject: Re: AW: Very basic question about Archive logs
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: "Subramanian,Ramachandran" <ramachandran.subramanian@alte-leipziger.de>,
  "David G. Johnston" <david.g.johnston@gmail.com>
Cc: "pgsql-novice@lists.postgresql.org" <pgsql-novice@lists.postgresql.org>
Date: Thu, 26 Mar 2026 14:00:36 +0100
In-Reply-To: <5604c5c610e44a0d97daceaf5f1ae01f@alte-leipziger.de>
References: <dee8d2457c034cddb7ab7021e1caf3f7@alte-leipziger.de>
	 <CAKFQuwZwA=sq7xeLV032QJCzZ99NQnq8SNg4a6Y_hBxN_D=PmA@mail.gmail.com>
	 <5604c5c610e44a0d97daceaf5f1ae01f@alte-leipziger.de>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.58.3 (3.58.3-1.fc43) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/8177cb9bc419f96f09302c50eb71d92723e01b9e.camel%40cybertec.at>
Precedence: bulk

On Thu, 2026-03-26 at 12:38 +0000, Subramanian,Ramachandran wrote:
> Just a theoretical question to better my understanding. If I take a backu=
p
> on Monday at 0100 Hours and never make any further Backups but hold on to
> all the Logs ( Active and Archive),
> I could theoretically go to any PIT =C2=A0between Monday 0100 Hours and N=
OW?

Yes.  That is, if the backup *ended* at 1:00 a.m.

> If I apply the base backup from Monday 0100 , how does Postgres know whic=
h
> Archive Logs need to be applied from it=E2=80=99s current state to get to=
 the PIT.
> Obviously when the basebackup was taken, the now needed logs did not exis=
t.
> So this information cannot have been recorded during the backup.

PostgreSQL knows where to start, because that is in the backup_label file.
If recovery reaches the end of the current WAL segment, and we have not
yet reached the recovery target, PostgreSQL will restore the next WAL segme=
nt
and continue recovery (actually, PostgreSQL prefetches WAL segments, but th=
at
doesn't affect the principle).
Every commit record in the WAL has the commit timestamp, and that is how
PostgreSQL can determine if it has reached the recovery target (if
recovery_target_time was used).

Yours,
Laurenz Albe