From owner-freebsd-questions@FreeBSD.ORG Tue Dec 20 17:29:38 2005 Return-Path: X-Original-To: freebsd-questions@freebsd.org Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 24FA916A41F for ; Tue, 20 Dec 2005 17:29:38 +0000 (GMT) (envelope-from neal@rigney.org) Received: from zinc.nocdirect.com (zinc.nocdirect.com [69.73.148.119]) by mx1.FreeBSD.org (Postfix) with ESMTP id AF2A543D45 for ; Tue, 20 Dec 2005 17:29:37 +0000 (GMT) (envelope-from neal@rigney.org) Received: from jrigney by zinc.nocdirect.com with local (Exim 4.52) id 1EolId-00027g-6y for freebsd-questions@freebsd.org; Tue, 20 Dec 2005 11:29:19 -0600 From: "Neal Rigney" To: freebsd-questions@freebsd.org X-Mailer: NeoMail 1.27 X-IPAddress: 192.156.110.32 MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Message-Id: Date: Tue, 20 Dec 2005 11:29:19 -0600 X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - zinc.nocdirect.com X-AntiAbuse: Original Domain - freebsd.org X-AntiAbuse: Originator/Caller UID/GID - [32116 2045] / [47 12] X-AntiAbuse: Sender Address Domain - rigney.org X-Source: X-Source-Args: X-Source-Dir: Subject: Occasional console freezing in 6.0-stable X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 20 Dec 2005 17:29:38 -0000 I previously sent out a query about a graid3 volume causing my system to lock hard, but there's now a new wrinkle: I removed the graid3 volume from my system, and am attempting to copy data off to a NFS mount. The NFS mount is connected via a GigE connection. The GigE card is a re(4)-based card, and seems to work fine in small packet mode. I was running in jumbo-frame mode last night, and received a console message a bout a frame that was too large - around 2100 bytes. I had used ifconfig to configure the mtu of the re card to 7644, the allowed value according to the re(4) manpage. >From the point of this message onward, the NFS mount would not work -- I got repeated "server not responding" messages. My console at this point still worked, and the two machines could still talk across the GigE link (ping and ssh both worked fine). I bounced the NFS server (not the machine, just the server) and reconfigured the cards for a 1500 mtu, thinking I was still getting frame errors under NFS and just wasn't seeing them. I then checked if I could see the NFS mount. Doing this resulted in the expected hang of my process. However, attempting to change to a new virtual console yielded only a beep from the machine, with no actual change of the console. Further attempts resulted in no beep and no console change. On a hunch, I used ssh to connect to the NFS client machine. It was up fine, with no load. The cpio process I was using to copy data was still there, hung on IO as I would expect from a dead NFS mount. Thinking that the server was at least safe even if it wasn't behaving really well, I attempted one more access of the NFS mount. This one, however, resulted in a complete hang of the system. No IP response, no keyboard response. At this point my wife has given up on having access to her media files for a few more days. I'm willing to experiment (I've got backups made). What can I do to help resolve this problem? I've got plenty of FreeBSD experience, but nothing since around 4.7 in the kernel world. --