Tom Lane wrote:
> Not directly, but if you use the contrib/pgstattuple module you can get
> readings on the numbers of committed and uncommitted tuples in the
> table. The rate at which the uncommitted-tuples count increases would
> tell you how fast the update is proceeding. (You should probably not
> assume that you started with zero uncommitted tuples, unless you know
> you'd vacuumed the table just beforehand.)
Thanks, Tom.
I've run it, and I see the dead_tuple_count, but I don't see an
uncommitted count. The dead_tuple_count seems to oscillate, too -- it
increases slowly, but sometimes decreases immediately after. Is that
normal?
I don't see an uncommitted tuples count, though.
--Colin McGuigan
