> Rod Taylor <pg@rbt.ca> writes:
> > Is there any chance it is related to the 8.0 problem I reported on
> > Wednesday?
>
> Too soon to tell ... though one would like to think we don't have more
> than one bug in that area ;-).
>
> If either of you can come up with even a low-probability test case,
> it'd be a great help.
>
> regards, tom lane
I spend several hours yesterday trying all different manner of things
and came up with zero. Eventually I got to the point where I felt
like I was wasting my time. ISTM there is likely some type of trigger
which will be difficult to catch via simulation. Our production
servers are running dual opterons and my dev box isn't, so it's
impossible for me to to a pefect simiulation :(.
While I was initially extremely concerned about this, an internal
audit confirmed there was no actual data loss...the proper row
appeared to be visible in all cases, so I classify this as more of
correctible annoyance, fwiw.
Talked with Rod a bit off-list and the best course of action is to
implement a nightly process to check for things and work from there.I'll keep you posted.
Merlin
