> Also, that patch doesn't address the VACUUM issue at all. And
> using a small fixed size ring with scans that do updates can
> be devastating. I'm experimenting with different ring sizes
> for COPY at the moment. Too small ring leads to a lot of WAL
> flushes, it's basically the same problem we have with VACUUM
> in CVS HEAD.
My first take on that would be to simply abandon any dirty (and actually
also any still pinned) buffer from the ring and replace the ring slot
with a buffer from the freelist.
If the freelist is empty and LSN allows writing the buffer, write it
(and maybe try to group these).
If the LSN does not allow the write, replace the slot with a buffer from
LRU.
Andreas