On Tue, Apr 20, 2021 at 01:38:48PM +0530, Bharath Rupireddy wrote:
> I think it comes under the category "Resource Usage / Asynchronous
> Behavior", so what pg_settings showing is correct. ISTM that we need
> to correct the docs, attached a patch for that.
This got introduced in e5253fd. My first impression was that this
had better be a developer option, but Thomas is mentioning an extra
reason why this category may not be a good fit:
https://www.postgresql.org/message-id/CAEepm=3G1-SKd2qKN-3uen=Xvyi-OxAVg9RAwqWDH-KZWuGqNA@mail.gmail.com
And Robert has moved that at the end to its GUC section without
addressing the docs:
https://www.postgresql.org/message-id/CA+Tgmoa=AxuLKo2f1AKwLmvFoo3rr3j+SYHaJ5GVyx1PhhOJ0Q@mail.gmail.com
So I agree that your patch is adapted, even postgresql.conf.sample
gets that right. Something that your patch makes worse is the
alphabetical order of the parameters listed in this section
(backend_flush_after can be also blamed here), so I'll go reorder this
sub-area a bit while on it, except if somebody objects.
--
Michael