Re: CLUSTER and MVCC

Heikki Linnakangas <heikki(at)enterprisedb(dot)com> writes:
> I'm thinking of keeping an in-memory mapping of old and new tids of
> updated tuples while clustering, instead. That means that cluster
> requires a little bit of memory for each RECENTLY_DEAD updated tuple. In
> the worst case that means that you run out of memory if there's too many
> of those in the table, but I doubt that's going to be a problem in practice.

That is more or less isomorphic to what VACUUM FULL does. While people
have complained about VACUUM FULL's memory usage on occasion, just at
the moment I feel that the main problem with it is complexity. If we
still haven't gotten all the bugs out of VACUUM FULL after more than
eight years of work on it, what are the odds that we can make CLUSTER
do it right the first time?

regards, tom lane