Harry Ambrose <harry.ambrose@gmail.com> writes:
> Not sure whether its relevant or not, however upon adding an ANALYSE before
> the second vacuum the issue has not presented when testing. I have managed
> 95 cycles thus far.
I'm still unable to reproduce :-( --- I ran about two dozen cycles
overnight with no sign of trouble. This time I was using a master/slave
pair with the test database in a non-default partition, so neither of
those aspects seem to be key after all.
I suspect the reason for it being so hard to reproduce is that there's
a timing window involved. But that doesn't offer much to go on in
terms of being able to make a more reproducible case.
regards, tom lane
