Bruce Momjian <bruce@momjian.us> writes:
> I thought the test was setting up a configuration that would never be
> used by normal servers. Is that false?
*If* we made it disable autovac before starting pg_upgrade,
then that would be a process not used by normal users.
I don't care whether pg_upgrade disables autovac during its
run; that's not what's at issue here.
regards, tom lane