Re: Is it correct to update db state in control file as "shutting down" during end-of-recovery checkpoint?

At Sat, 26 Feb 2022 12:11:15 +0900, Michael Paquier <michael@paquier.xyz> wrote in 
> On Fri, Feb 25, 2022 at 01:09:53PM -0800, Nathan Bossart wrote:
> > This one has been quiet for a while.  Should we mark it as
> > returned-with-feedback?
> 
> Yes, that's my feeling and I got cold feet about this change.  This
> patch would bring some extra visibility for something that's not
> incorrect either on HEAD, as end-of-recovery checkpoints are the same
> things as shutdown checkpoints.  And there is an extra argument where
> back-patching would become a bit more tricky in an area that's already
> a lot sensitive.

That sounds like we should reject the patch as we don't agree to its
objective.  If someday end-of-recovery checkpoints functionally
diverge from shutdown checkpoints but leave (somehow) the transition
alone, we may visit this again but it would be another proposal.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center