Pavan Deolasee <pavan.deolasee@gmail.com> writes:
> What if we remember the buffers as seen by count_nondeletable_pages() and
> then just discard those specific buffers instead of scanning the entire
> shared_buffers again?
That's an idea.
> Surely we revisit all to-be-truncated blocks before
> actual truncation. So we already know which buffers to discard. And we're
> holding exclusive lock at that point, so nothing can change underneath. Of
> course, we can't really remember a large number of buffers, so we can do
> this in small chunks.
Hm? We're deleting the last N consecutive blocks, so it seems like we
just need to think in terms of clearing that range. I think this can
just be a local logic change inside DropRelFileNodeBuffers().
You could optimize it fairly easily with some heuristic that compares
N to sizeof shared buffers; if it's too large a fraction, the existing
implementation will be cheaper than a bunch of hashtable probes.
regards, tom lane