On 4/13/23 11:32 AM, Jonathan S. Katz wrote:
> On 4/12/23 11:34 PM, Amit Kapila wrote:
>> On Tue, Apr 11, 2023 at 11:50 PM Andres Freund <andres@anarazel.de>
>> +1 to do one of the above. I think there is a good chance that
>> somebody might be doing more harm by using it so removing this
>> shouldn't be a problem. Personally, I have not heard of people using
>> it but OTOH it is difficult to predict so giving some time is also not
>> a bad idea.
>>
>> Do others have any opinion/suggestion on this matter?
>
> I need a bit more time to study this before formulating an opinion on
> whether we should remove it for v16. In any case, I'm not against
> documentation.
(didn't need too much more time).
[RMT hat]
+1 for removing.
I looked at some data and it doesn't seem like vacuum_defer_cleanup_age
is used in any significant way, whereas hot_standby_feedback is much
more widely used. Given this, and all the problems + arguments made in
the thread, we should just get rid of it for v16.
There are cases where we should deprecate before removing, but I don't
think this one based upon usage and having a better alternative.
Per [1] it does sound like we can make some improvements to
hot_standby_feedback, but those can wait to v17.
We should probably set $DATE to finish this, too. I don't think it's a
rush, but we should give enough time before Beta 1.
Jonathan
[1]
https://www.postgresql.org/message-id/20230317230930.nhsgk3qfk7f4axls%40awork3.anarazel.de