Re: pgbench regression test failure

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr>
Cc: PostgreSQL Developers <pgsql-hackers(at)postgreSQL(dot)org>
Subject: Re: pgbench regression test failure
Date: 2017-09-12 18:12:16
Message-ID: 6751.1505239936@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr> writes:
> By definition, parallelism induces non determinism. When I put 2 seconds,
> the intention was that I would get a non empty trace with a "every second"
> aggregation. I would rather take a longer test rather than allowing an
> empty file: the point is to check that something is generated, but
> avoiding a longer test is desirable. So I would suggest to stick to
> between 1 and 3, and if it fails then maybe add one second...

That's a losing game. You can't ever guarantee that N seconds is
enough for slow, heavily loaded machines, and cranking up N just
penalizes developers who are testing under normal circumstances.

I have a serious, serious dislike for tests that seem to work until
they're run on a heavily loaded machine. So unless there is some
reason why pgbench is *guaranteed* to run at least one transaction
per thread, I'd rather the test not assume that.

I would not necessarily object to doing something in the code that
would guarantee that, though.

regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2017-09-12 18:28:51 Re: domain type smashing is expensive
Previous Message Fabien COELHO 2017-09-12 18:00:56 Re: pgbench regression test failure