From owner-freebsd-perl@FreeBSD.ORG  Sat Oct  1 19:58:50 2005
Return-Path: <owner-freebsd-perl@FreeBSD.ORG>
X-Original-To: perl@FreeBSD.org
Delivered-To: freebsd-perl@FreeBSD.ORG
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 05D2316A41F;
	Sat,  1 Oct 2005 19:58:50 +0000 (GMT)
	(envelope-from zefram@zefram.net)
Received: from dante.zefram.net (dante.zefram.net [66.92.77.109])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 069DB43D46;
	Sat,  1 Oct 2005 19:58:48 +0000 (GMT)
	(envelope-from zefram@zefram.net)
Received: by dante.zefram.net (Postfix, from userid 1000)
	id A4B5813A; Sat,  1 Oct 2005 15:58:49 -0400 (EDT)
Received: from localhost (localhost [127.0.0.1])
	by dante.zefram.net (Postfix) with ESMTP id 9B0DB94;
	Sat,  1 Oct 2005 15:58:49 -0400 (EDT)
Date: Sat, 1 Oct 2005 15:58:49 -0400 (EDT)
From: John Gillis <zefram@zefram.net>
To: Anton Berezin <tobez@FreeBSD.org>
In-Reply-To: <200509282107.j8SL7oGO005442@freefall.freebsd.org>
Message-ID: <20051001154734.E17540@dante.zefram.net>
References: <200509282107.j8SL7oGO005442@freefall.freebsd.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Cc: perl@FreeBSD.org
Subject: Re: ports/86687: Perl ithreads coredump
X-BeenThere: freebsd-perl@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: maintainer of a number of perl-related ports
	<freebsd-perl.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-perl>,
	<mailto:freebsd-perl-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-perl>
List-Post: <mailto:freebsd-perl@freebsd.org>
List-Help: <mailto:freebsd-perl-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-perl>,
	<mailto:freebsd-perl-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Sat, 01 Oct 2005 19:58:50 -0000

	Here's a sample script that creates the same core dump. It creates
50 threads that look up 1000 hostnames from a database of unique
hostnames. They are all unique. It then just runs gethostbyname.
	The actual script uses IO::Socket::INET... but I didn't want to
connect to so many hosts at once.. Without further adieu, here's the
script:

#!/usr/bin/perl -w

use threads;
use Socket;

$SIG{INT}                = \&catch_die;

my @threads;
my $shutdown :shared = 0;

for (1 .. 50) {
  shift @threads, threads->new(\&thr);
}

sub thr {
  my $i   = 0;
  my $dbi = DBI->connect("DBI:mysql:database=spider","xxx","xxx") or
die "Unable to connect to MySQL: $!\n";

  my $sth = $dbi->prepare("SELECT id, host FROM site LIMIT " .
(threads->tid * 1000) . ", 1000");
  $sth->execute;

  return if !$sth->rows;

  while ($shutdown == 0 && $i < 999) {
    my @row   = $sth->fetchrow();
    my $host  = $row[1];
    my $iaddr = gethostbyname($host) || print "Host $host not found.";
    $i++;
  }
}

while ($shutdown == 0) {
  sleep(5);
}

print "Waiting for threads\n";

foreach $thread (@threads) {
  $thread->join;
}

sub catch_die {
  # Shit. Gotta clean our act up.
  print "Closing once current page has been indexed.\n";
  $shutdown = 1;
}