"Patrick Hatcher" <PHatcher@macys.com> writes:
> Should it take over 8 hours to do a vacuum full analyze on a 15 million
> record table? I have no idea when, if ever, this user has done a vacuum
> full on this machine.
With no more context than that, the only possible answer is "who knows?"
You've given us no clue about the hardware you're doing this on, so
the single datapoint that there are 15 million live rows doesn't let
anyone calibrate anything.
> I'm afraid to cancel the process for fear of corrupting the table.
FWIW, cancelling a VACUUM is perfectly safe (were it not, a power loss
during VACUUM would be fatal). But now that you've got 8 hours
invested, I'd counsel letting it run ... if you cancel, the next attempt
will take even longer.
regards, tom lane