I wrote:
> Daniel Wood <dwood@salesforce.com> writes:
>> Does the original version of my stress test not repro the problem on 9.2?
> [ tries it ... ] No, it doesn't, or at least the MTBF is a couple orders
> of magnitude better than on 9.3.
Oh, of course: the reason the test doesn't fail as given on 9.2 is that
9.2 doesn't have a lock_timeout parameter. (I missed the complaints about
this in the blizzard of other noise the test generates :-(.) The timeout
is critical to exposing the bug because it needs failed lock acquisitions.
Probably unpatched 9.2 would fall over if you used statement_timeout
instead.
regards, tom lane