Thomas Munro wrote:
> I can't help thinking there must be a different way to do this that
> takes advantage of the fact that multixacts are often created by
> copying all the members of an existing multixact and adding one new
> one, so that there is a lot of duplication and churn (at least when
> you have a workload that generates bigger multixacts, due to the
> O(n^2) process of building them up xid by xid).
Yeah, Simon expressed the same thought to me some months ago, and I gave
it some think-time (but not at lot of it TBH). I didn't see any way to
make it workable.
Normally, lockers go away reasonably quickly, so some of the original
members of the multixact are disappearing all the time. Maybe one way
would be to re-use a multixact you have in your local cache, as long as
the only difference with the multixact you want is some locker
transaction(s) that have already ended. Not sure how you would manage
the cache, though.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services