Re: RFE: Make statistics robust for unplanned events

On Wed, Apr 21, 2021 at 5:05 PM Magnus Hagander <magnus(at)hagander(dot)net> wrote:

>
> > Right. I think the other question is how often does this happen in
> > practice - if your instance crashes often enough to make this an issue,
> > then there are probably bigger issues.
>
> Agreed.
>
> I think the bigger problem there is replication failover, but that's
> also a different issue (keeping the statistics from the *standby*
> wouldn't help you much there, you'd need to replicate it from the
> primary).
>
> --
> Magnus Hagander
> Me: https://www.hagander.net/
> Work: https://www.redpill-linpro.com/
>
>
The report that I've received regarding this RFE has been triggered by
experiencing issues with long term deployments in a large scale industrial
environment. The point of this RFE is to be protected against those issues
in the future. While this doesn't seem to be a very frequent occurrence, I
wouldn't consider this a corner case not being worth attention.

If there is an expectation for the performance loss to be less of a problem
in the future, would it make sense to make this an opt-in feature until
then?

--
Patrik Novotný
Associate Software Engineer
Red Hat
panovotn(at)redhat(dot)com