On Tue, 22 Nov 2022 at 16:53, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Simon Riggs <simon.riggs@enterprisedb.com> writes:
> > On Tue, 22 Nov 2022 at 16:28, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >> If we do those things, do we need a wasted-work counter at all?
>
> > The wasted work counter works well to respond to heavy read-only
> > traffic and also avoids wasted compressions for write-heavy workloads.
> > So I still like it the best.
>
> This argument presumes that maintenance of the counter is free,
> which it surely is not. I don't know how bad contention on that
> atomically-updated variable could get, but it seems like it could
> be an issue when lots of processes are acquiring snapshots.
I understand. I was assuming that you and Andres liked that approach.
In the absence of that approach, falling back to a counter that
compresses every N xids would be best, in addition to the two new
forced compression events.
--
Simon Riggs http://www.EnterpriseDB.com/