On Mon, Mar 30, 2020 at 09:20:15AM -0700, Mark Dilger wrote:
> I have not been working on this issue lately, but as I recall, my
> concern was that changing the behavior of autovacuum could introduce
> regressions for some users, so we should be careful to get it right
> before we rush to release anything. It didn't seem like the
> proposed changes took enough into account. But that's clearly a
> judgement call, having to do with how cautious any particular person
> thinks we should be. I don't feel strongly enough to stand in the
> way if the general concensus is that this is a good enough
> implementation.
Echoing with what has been already mentioned on this thread, I think
that autovacuum scheduling is a hard problem, and I would be rather
scared to change by default a behavior that has proved to work in some
cases, but could potentially doom others. I have an idea though: we
could make the scheduling behavior of autovacuum optional.
Anyway, the thread has stalled for a couple of months now, and we
don't have a clear consensus about this approach, so I am marking this
thread as returned with feedback.
--
Michael