From: Tom Lane <tgl@sss.pgh.pa.us>
To: Noah Misch <noah@leadboat.com>
cc: pgsql-hackers@postgresql.org
Subject: Re: pg_regress: promptly detect failed postmaster startup
In-reply-to: <20181231172922.GA199150@gust.leadboat.com>
References: <20181231172922.GA199150@gust.leadboat.com>
Comments: In-reply-to Noah Misch <noah@leadboat.com>
	message dated "Mon, 31 Dec 2018 12:29:22 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <19453.1546282291.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 31 Dec 2018 13:51:31 -0500
Message-ID: <19454.1546282291@sss.pgh.pa.us>
Precedence: bulk

Noah Misch <noah@leadboat.com> writes:
> When "make check TEMP_CONFIG=3D<(echo break_me=3Don)" spawns a postmaste=
r that
> fails startup, we detect that with "pg_regress: postmaster did not respo=
nd
> within 60 seconds".  pg_regress has a kill(postmaster_pid, 0) intended t=
o
> detect this case faster.  Since kill(ZOMBIE-PID, 0) succeeds[1], that te=
st is
> ineffective.

Ooops.

> The fix, attached, is to instead test waitpid(), like pg_ctl's
> wait_for_postmaster() does.

+1.  This leaves postmaster_pid as a dangling pointer, but since
we just exit immediately, that seems fine.  (If we continued, and
arrived at the "kill(postmaster_pid, SIGKILL)" below, it would not
be fine.)

			regards, tom lane