On Tue, Jun 25, 2024 at 09:22:48AM +0900, Kyotaro Horiguchi wrote:
> Is it something like this? I forgot the reason why this was excluded
> from the vinal version, though.
>
> https://www.postgresql.org/message-id/20201001.090722.322196928507744460.horikyota.ntt%40gmail.com
That's a bit different than the gc_count presented in 0004, because
there is no relationship to autovacuum with the cleanup of the
entries, still there is some overlap in the thought process: this
suggestion just makes sure that any backend holding local references
to a shared entry with a different age does not attempt to manipulate
it after switching locally to a dropped state.
--
Michael