From owner-freebsd-hardware Tue Jun 2 07:12:39 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id HAA01735 for freebsd-hardware-outgoing; Tue, 2 Jun 1998 07:12:39 -0700 (PDT) (envelope-from owner-freebsd-hardware@FreeBSD.ORG) Received: from russian-caravan.cloud9.net (russian-caravan.cloud9.net [168.100.1.4]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id HAA01729 for ; Tue, 2 Jun 1998 07:12:35 -0700 (PDT) (envelope-from scottd@cloud9.net) Received: from earl-grey.cloud9.net (xkfs4sz+Nrf2SFd6Av41SBiEoiaha3IR@earl-grey.cloud9.net [168.100.1.1]) by russian-caravan.cloud9.net (8.8.8/8.8.8) with SMTP id KAA22037; Tue, 2 Jun 1998 10:12:31 -0400 (EDT) Date: Tue, 2 Jun 1998 10:12:30 -0400 (EDT) From: Scott Drassinower Reply-To: Scott Drassinower To: David Greenman cc: freebsd-hardware@FreeBSD.ORG Subject: Re: Intel EtherExpress 100+ problems In-Reply-To: <199806020432.VAA12291@implode.root.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hardware@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I will try forcing the two machines to use 100/half in the switch. The interface counters for the Cisco really don't show any major errors or drops for those interfaces (there were a handful of input errors for the server, none for other machines). I don't remember whose cable it is, but I think it is Enhanced Cat 5, and the cables are all about 6 feet. I could try different cable, but I'm not sure that would change the situation. netstat -i on the client has no interface errors, netstat -i on the server has one output error but no input errors. I might be able to get a hold of another 100 megabit switch, perhaps an Intel, and we can see if the problem goes away or if it's the same. I'm wondering if there are other people running 100 megabit that might be able to duplicate this problem. -- Scott M. Drassinower scottd@cloud9.net Cloud 9 Consulting, Inc. White Plains, NY +1 914 696-4000 http://www.cloud9.net On Mon, 1 Jun 1998, David Greenman wrote: > >So the problem that I'm having with NFS will likely occur with a 3Com or > >SMC card as well? There is no workaround at all for this, just deal with > >messed up nfs or run at 10 megabits? > > I was only suggesting that our NFS has problems in some corner cases. > Usually this only shows up when there are multiple clients writing to the > same file, and the usual result is file corruption, not a client/server > wedge. Our NFS also doesn't recover properly from certain kinds of failures. > All of these issues are being looked at in -current and there have been a > few fixes that have trickled back to -stable...not that I think those will > fix your problem, however. I think your troubles are at the link level and > are symptoms of packet loss. > Try setting both sides to forced 100/half. I've had some compatiblity > problems with Cisco router-switches when connected to other vendor switches > in 100/full. For some reason switching to half duplex causes the problems > to go away (and it's not actually a duplex problem as that was verified to > be correct on both ends...weird). > I haven't seen the problems you are having with NFS and the Cisco/Intel > combination. It sounds like there is packet loss in some cases which is > causing NFS clients to wedge. Are you sure that your cable is up to spec? > Are there any interface errors on either side? > > -DG > > David Greenman > Co-founder/Principal Architect, The FreeBSD Project > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message