From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [RFC] random SYN drops causing connect() delays Date: Wed, 28 Apr 2010 07:52:46 +0200 Message-ID: <1272433966.2343.34.camel@edumazet-laptop> References: <20100412080633.GA27418@bombadil.infradead.org> <20100412083903.GA19763@bombadil.infradead.org> <20100427.185602.221594952.davem@davemloft.net> <20100428044443.GL19763@bombadil.infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: David Miller , netdev@vger.kernel.org To: Thomas Graf Return-path: Received: from mail-bw0-f219.google.com ([209.85.218.219]:57982 "EHLO mail-bw0-f219.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752053Ab0D1Fww (ORCPT ); Wed, 28 Apr 2010 01:52:52 -0400 Received: by bwz19 with SMTP id 19so196182bwz.21 for ; Tue, 27 Apr 2010 22:52:51 -0700 (PDT) In-Reply-To: <20100428044443.GL19763@bombadil.infradead.org> Sender: netdev-owner@vger.kernel.org List-ID: Le mercredi 28 avril 2010 =C3=A0 00:44 -0400, Thomas Graf a =C3=A9crit = : > On Tue, Apr 27, 2010 at 06:56:02PM -0700, David Miller wrote: > > I can't reproduce on my system even with sched_child_runs_first set= to '1'. > >=20 > > Are you running identd or something like that which intercepts the = connections > > to port 22 before 'sshd' actually gets it? >=20 > No, sshd is listening on the port directly. >=20 > I am having more difficulties to reproduce the issue with the latest = git tree. It > still happens but I sometimes have to run the test several times to g= et a single > delayed connection attempt. On machine/kernel reproducing the behavior, please send grep . /proc/sys/net/ipv4/*