Re: Instability in TRUNCATE regression test

Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
> Tom Lane wrote:
>> Buildfarm member platypus is showing a regression failure that I'm
>> surprised we have not seen before:
>> http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=platypus&dt=2006-06-28%2014:05:01

> If this were a significant risk wouldn't we have seen many such failures
> before now?

mongoose just failed with almost the exact same symptoms:

http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=mongoose&dt=2006-06-28%2021:30:02

It's not quite the same diffs, which is unsurprising given the presumed
mechanism behind the failure, but that probably shoots down the "add
another expected file" response.

I imagine some recent change has made the probability of this behavior
much higher than it was before; perhaps there's more pg_constraint
update traffic in concurrent tests? Anyway, it's now up to "must fix"
in my estimation. I'll look into the new-pg_constraint-index idea.

I think someone should also take a hard look at that idea of introducing
more timing variability into the parallel tests. Any volunteers?

regards, tom lane