On Thu, Feb 09, 2023 at 02:51:14PM -0800, Nathan Bossart wrote:
> Hm...  We might be able to use WaitLatch() for some of these.

Perhaps less than you think as a bit of work has been done in the last
few years to reduce the gap and make such code paths more responsive,
still I would not be surprised to find a couple of these..

I would let the portions of the code related to things like
pre_auth_delay or XLOG_REPLAY_DELAY as they are, though, without an
extra pg_Xsleep() implementation.
--
Michael

Attachment: signature.asc
Description: PGP signature

Reply via email to