Tom Lane wrote:
> "Heikki Linnakangas" <heikki@enterprisedb.com> writes:
>> Bruce Momjian wrote:
>>> This might be a simplistic question but if the page is +90% full and
>>> there is a long-lived transaction, isn't Postgres going to try pruning
>>> on each page read access?
>
>> Yes :(
>
> It shouldn't, though --- the hint bit should get cleared on the first
> try.
Think so? That would mean that you only get a single chance to prune
after an update. Not sure how big a problem that is, but I do feel that
would make HOT a lot less effective in some usage patterns.
The idea of having a "prunable xmin" in the page header sounds more and
more attractive to me...
> I think I probably broke something in the last round of revisions
> to heap_page_prune_opt, but haven't looked yet ...
Pavan's patch was like that as well; you don't clear the flag (or
rather, you set it again while pruning) until there's nothing left to
prune on the page.
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com