From owner-freebsd-net@FreeBSD.ORG Thu Jul 19 08:13:25 2007 Return-Path: X-Original-To: freebsd-net@freebsd.org Delivered-To: freebsd-net@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 1C94316A402 for ; Thu, 19 Jul 2007 08:13:25 +0000 (UTC) (envelope-from andre@freebsd.org) Received: from c00l3r.networx.ch (c00l3r.networx.ch [62.48.2.2]) by mx1.freebsd.org (Postfix) with ESMTP id 8259213C481 for ; Thu, 19 Jul 2007 08:13:24 +0000 (UTC) (envelope-from andre@freebsd.org) Received: (qmail 86828 invoked from network); 19 Jul 2007 08:09:55 -0000 Received: from c00l3r.networx.ch (HELO [127.0.0.1]) ([62.48.2.2]) (envelope-sender ) by c00l3r.networx.ch (qmail-ldap-1.03) with SMTP for ; 19 Jul 2007 08:09:55 -0000 Message-ID: <469F1D32.2040700@freebsd.org> Date: Thu, 19 Jul 2007 10:13:38 +0200 From: Andre Oppermann User-Agent: Thunderbird 1.5.0.12 (Windows/20070509) MIME-Version: 1.0 To: Jack Vogel References: <2a41acea0707181152r4c662b00tf1be7b6b4dce74f1@mail.gmail.com> <2a41acea0707181542q23d36733o339448ad86f480cb@mail.gmail.com> In-Reply-To: <2a41acea0707181542q23d36733o339448ad86f480cb@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: freebsd-net , FreeBSD Current Subject: Re: 10G and socket alloc failure X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jul 2007 08:13:25 -0000 Jack Vogel wrote: > While testing out the ixgbe driver we've observed this failure in the stack > code, here is the info: > > The test engineer is using iperf, typically with 16 threads. If the > driver is using > either legacy or MSI interrupts we will see broken pipes, in dmesg its > due to sonewconn() failing in syncache_socket(). > > Whats interesting is that when I have multiple RX queues configured and > using MSI/X this doesnt happen, at least not very often. > > It does not seem to hurt performance horribly, iperf just spawns another > thread, but I was wondering if there is some underlying tuneable I dont > know about that would stop this from happening?? > > And any theory about why it doesnt happen with multiple queues? Do you see any messages in syslog regarding syncache? -- Andre