Jeremy Yoder <jyoder@fame.com> writes:
> I compiled with zlib 1.2.1 and can't reproduce the error, but that
> probably has nothing to do with it. It looks like there's a sleep right
> before the test that failed. Perhaps it didn't sleep long enough?
Much more likely this indicates that the stats collector failed to start
up. The reason the test is there is that we've seen this happen on
systems with strangely-configured networking. Look at the postmaster's
log file (stderr output or syslog, depending on how you have it
configured) to see if there's any bleating about not being able to bind
to the stats collection socket, or errors along that line. They'd be
right near the top, during postmaster startup.
In any case, showing only the output "stats ... FAILED" is of little use
in diagnosing the reason for a test failure. You have to look at the
actual regression.diffs file if you want to do more than guess.
regards, tom lane
