Horton

Re: please update ps display for recovery checkpoint

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: "k.jamison@fujitsu.com" <k.jamison@fujitsu.com>
Cc: 'Justin Pryzby' <pryzby@telsasoft.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-08-20T08:09:05Z
Lists: pgsql-hackers 
On Wed, Aug 19, 2020 at 12:20:50AM +0000, k.jamison@fujitsu.com wrote:
> On Wednesday, August 19, 2020 7:53 AM (GMT+9), Justin Pryzby wrote: 
>> During crash recovery, the server writes this to log:
>> Please change to say "recovery checkpoint" or similar, as I mentioned here.
>> https://www.postgresql.org/message-id/20200118201111.GP26045@telsasoft.c
>> om
> 
> Yes, I agree that it is helpful to tell users about that.

That could be helpful.  Wouldn't it be better to use "end-of-recovery
checkpoint" instead?  That's the common wording in the code comments.

I don't see the point of patch 0002.  In the same paragraph, we
already know that this applies to any checkpoints.

> About 0003 patch, there are similar phrases in bgwriter_flush_after and 
> backend_flush_after. Should those be updated too?

Yep, makes sense.
--
Michael

Commits

  1. Provide recovery_init_sync_method=syncfs.

  2. Add some checkpoint/restartpoint status to ps display