On Mon, Jan 24, 2011 at 9:26 PM, Josh Berkus <josh(at)agliodbs(dot)com> wrote:
> It looks to me like autovacuum doesn't ever consider when HOT updates
> lead to page splits, and so require vacuuming. Or am I diagnosing it wrong?
I'm not sure what you mean by a page split. An update wherein the new
heap tuple won't fit on the same page as the existing heap tuple
should be treated as non-HOT. But nothing gets split in that case. I
think of a page split as an index event, and if these are HOT updates
there shouldn't be any index changes at all.
Can we see those stats again with n_tup_ins/upd/del?
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company