I wrote:
> Andres Freund <andres@anarazel.de> writes:
>> I think what might be happening is that the transactional stats updates get
>> reported by s2 *before* the non-transactional stats updates come in from
>> s1. I.e. the pgstat_report_stat() at the end of s2_commit_prepared_a does a
>> report, because the machine is slow enough for it to be "time to reports stats
>> again". Then s1 reports its non-transactional stats.
> Sounds plausible. And I left the test loop running, and it's now past
> 100 consecutive successes, so I think this change definitely "fixes" it.
In the light of morning, at least half of the parameter dependency is
now obvious: the problematic test case involves a prepared transaction,
so it fails completely with max_prepared_transactions = 0. The isolation
test harness masks that by matching against stats_1.out, but it's not
really a "success".
My numbers do still suggest that there's a weak dependence on
max_connections, but it's possible that that's a mirage. I did not run
enough test cycles to be able to say positively that that's a real effect
(and the machine's slow enough that I'm not excited about doing so).
regards, tom lane