Re: Improve pg_sync_replication_slots() to wait for primary to advance

On Fri, Aug 29, 2025 at 2:37 PM shveta malik <shveta.malik@gmail.com> wrote:
>
> On Fri, Aug 29, 2025 at 2:20 PM Ashutosh Bapat
> <ashutosh.bapat.oss@gmail.com> wrote:
> >
> > @@ -1276,7 +1331,7 @@ wait_for_slot_activity(bool some_slot_updated)
> >
> > The function is too cute to be useful. The code should be part of
> > ReplSlotSyncWorkerMain() just like other worker's main functions.
> >
>
> I was thinking we can retain wait_for_slot_activity() as this can even
> be invoked from API flow. See my comment# 2 in [1]

We want the SQL callable function to finish as fast as possible, and
make all the slots sync ready as fast as possible. So a shorter nap
time makes sense. We don't want to increase it per iteration. But sync
worker is a long running worker and can afford to wait longer. In fact
it should wait longer so as not to load the primary and the standby.
Given that the naptimes in both cases can not be controlled by the
same logic, I think it's better not to use the same function. Each of
them should have separate code for napping. That way the logic which
decides the nap time is closer to the code that naps making it more
readable.

--
Best Wishes,
Ashutosh Bapat