From owner-freebsd-fs@FreeBSD.ORG Thu Jun 10 11:46:16 2010 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id C31141065676 for ; Thu, 10 Jun 2010 11:46:16 +0000 (UTC) (envelope-from jdc@koitsu.dyndns.org) Received: from qmta07.emeryville.ca.mail.comcast.net (qmta07.emeryville.ca.mail.comcast.net [76.96.30.64]) by mx1.freebsd.org (Postfix) with ESMTP id A830F8FC1A for ; Thu, 10 Jun 2010 11:46:15 +0000 (UTC) Received: from omta01.emeryville.ca.mail.comcast.net ([76.96.30.11]) by qmta07.emeryville.ca.mail.comcast.net with comcast id UBbX1e0040EPchoA7BmFsj; Thu, 10 Jun 2010 11:46:15 +0000 Received: from koitsu.dyndns.org ([98.248.46.159]) by omta01.emeryville.ca.mail.comcast.net with comcast id UBmE1e00A3S48mS8MBmEYg; Thu, 10 Jun 2010 11:46:15 +0000 Received: by icarus.home.lan (Postfix, from userid 1000) id 692D99B418; Thu, 10 Jun 2010 04:46:14 -0700 (PDT) Date: Thu, 10 Jun 2010 04:46:14 -0700 From: Jeremy Chadwick To: Anders Nordby Message-ID: <20100610114614.GA71432@icarus.home.lan> References: <20100608083649.GA77452@fupp.net> <20100609122517.GA16231@fupp.net> <20100610111316.GB87243@fupp.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20100610111316.GB87243@fupp.net> User-Agent: Mutt/1.5.20 (2009-06-14) Cc: freebsd-fs@FreeBSD.org Subject: Re: Odd network issues on ZFS based NFS server X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 10 Jun 2010 11:46:16 -0000 On Thu, Jun 10, 2010 at 01:13:16PM +0200, Anders Nordby wrote: > On Wed, Jun 09, 2010 at 11:28:52AM -0400, Rick Macklem wrote: > > When you tried a different NIC, was a different type (ie. different > > chipset that uses a different device driver)? I suggested that not > > because I thought the hardware was broken but because I thought it > > might be related to the network interface's device driver and switching > > to a different device driver would isolate that possibility. > > Nope. I switched from NIC 1 to 2, and switched server to an identical > one. They both use bge NICs, a very common interface. I somehow doubt > this is related to the NIC or driver, I have many machines with the same > bge NIC (HP NC7782) that does not have any problems like this. This may not be the problem of course, but are they they *exact* same model and revision of NIC? pciconf -lvc on both boxes, and looking for the relevant bgeX interfaces, would determine that. I believe Rick was recommending you switch to another model of NIC that doesn't fall under the same driver, e.g. do you see this behaviour when using em(4). Also, can you provide uname -a output, or specifically the build date of the kernel. There have been bge(4) changes happening regularly throughout the lifetime of RELENG_8, including into the -PRERELEASE stage. > Mem: 25M Active, 147M Inact, 1021M Wired, 112K Cache, 316M Buf, 1780M Free > Swap: 6144M Total, 6144M Free > > The server doesn't have many connections, 16 in ESTABLISHED state. As > you can see from top, the server has 1780 MB free memory. Clarification: I believe it actually has 1927MB (147M Inact + 1780M Free) available. I've always understood top's "Free" field to mean "number/amount of pages which have never been touched/used since the kernel was started", while "Inact" to mean "number/amount of pages which have been touched/used but are not actively being used, this available for use". If someone more familiar with the VM and top could expand on this, that'd be helpful. -- | Jeremy Chadwick jdc@parodius.com | | Parodius Networking http://www.parodius.com/ | | UNIX Systems Administrator Mountain View, CA, USA | | Making life hard for others since 1977. PGP: 4BD6C0CB |