Mike Mascari <mascarm(at)mascari(dot)com> writes:
> Assuming a "normal" usage pattern, regular VACUUMing, and no
> instances of corrupted indexes, are there any scenarios in which one
> would need to REINDEX either user or system tables post 7.4?
Ideally not, but we'll have to wait for more field experience before we
really know whether the existing fix covers all "normal" usage patterns.
regards, tom lane