Re: Documentation patch for backup manifests in protocol.sgml - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: Documentation patch for backup manifests in protocol.sgml
Date
Msg-id 20200824075834.GA25747@paquier.xyz
Whole thread Raw
In response to Re: Documentation patch for backup manifests in protocol.sgml  (Bruce Momjian <bruce@momjian.us>)
Responses Re: Documentation patch for backup manifests in protocol.sgml  (Bruce Momjian <bruce@momjian.us>)
List pgsql-hackers
On Fri, Aug 21, 2020 at 06:03:32PM -0400, Bruce Momjian wrote:
> On Tue, Aug 18, 2020 at 02:41:09PM +0200, Bernd Helmle wrote:
>> protocol.sgml describes the protocol messages received by a BASE_BACKUP
>> streaming command, but doesn't tell anything about the additional
>> CopyResponse data message containing the contents of the backup
>> manifest (if requested) after having received the tar files. So i
>> propose the attached to give a little more detail in this paragraph.
>
> If someone can confirm this, I will apply it?  Magnus?

The reason why backup manifests are sent at the end of a base backup
is that they include the start and stop positions of the backup (see
caller of AddWALInfoToBackupManifest() in perform_base_backup()).
Once this is done, an extra CopyOutResponse message is indeed sent
within SendBackupManifest() in backup_manifest.c.

So confirmed.
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Tatsuro Yamada
Date:
Subject: Re: list of extended statistics on psql
Next
From: Michael Paquier
Date:
Subject: Re: Creating a function for exposing memory usage of backend process