On Wed, Nov 16, 2022 at 1:47 AM Bharath Rupireddy
<bharath.rupireddyforpostgres@gmail.com> wrote:
> That can be done, only if we can disable the timeout in another place
> when the StandbyMode is set to true in ReadRecord(), that is, after
> the standby server finishes crash recovery and enters standby mode.
Oh, interesting. I didn't realize that we would need to worry about that case.
> I'm attaching the v3 patch for further review. Please find the CF
> entry here - https://commitfest.postgresql.org/41/4012/.
I kind of dislike having to have logic for this in two places. Seems
like it could create future bugs.
How about the attached approach, instead? This way, the first time the
timer expires after we reach standby mode, we reactively disable it.
--
Robert Haas
EDB: http://www.enterprisedb.com