"PostgreSQL Bugs List" <pgsql-bugs(at)postgresql(dot)org> writes:
> At this point, the vacuum is *still* in progress (after 8 hours) and disc
> activity is exclusively the result of this vacuum, about 140 blocks/second
> with no disc queues. Is this likely to be a bug or just a scalability issue
> involving vacuum and the large object table?
vacuum full can take an awfully long time. It might be worth your while
to try a CLUSTER instead.
regards, tom lane