From owner-freebsd-questions@FreeBSD.ORG Fri Aug 18 13:06:53 2006 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 009C416A4DE for ; Fri, 18 Aug 2006 13:06:53 +0000 (UTC) (envelope-from mikhailg@webanoide.org) Received: from overlord.navalradio.cl (overlord.navalradio.cl [201.236.67.146]) by mx1.FreeBSD.org (Postfix) with ESMTP id 30F6C43D4C for ; Fri, 18 Aug 2006 13:06:51 +0000 (GMT) (envelope-from mikhailg@webanoide.org) Received: from [192.168.0.4] (ppp108-110.static.internode.on.net [150.101.108.110]) (authenticated bits=0) by overlord.navalradio.cl (8.13.6/8.13.6) with ESMTP id k7ID6o3Q001342 for ; Fri, 18 Aug 2006 09:06:52 -0400 (CLT) (envelope-from mikhailg@webanoide.org) Message-ID: <44E5BB5B.9010406@webanoide.org> Date: Fri, 18 Aug 2006 23:06:35 +1000 From: Mikhail Goriachev Organization: Webanoide User-Agent: Thunderbird 1.5.0.5 (Macintosh/20060719) MIME-Version: 1.0 To: freebsd-questions@freebsd.org Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Status: No, score=-100.0 required=5.0 tests=USER_IN_WHITELIST autolearn=failed version=3.1.4 X-Spam-Checker-Version: SpamAssassin 3.1.4 (2006-07-25) on overlord.navalradio.cl Subject: :::. P4 HTT and DUMP: Corrupted MAC on input. 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: Fri, 18 Aug 2006 13:06:53 -0000 Hi, I've got the following problem. A few servers, periodically dump(8) through ssh to a central server. That central server comes with a P4 HTT (single core - no frills) processor. A few weeks ago I decided to activate hyperthreading and add SMP capabilities to it. Just for the sake of it. Before those modifications, all servers were dumping without any problems for months, literally. However, now I get random disconnects. It may disconnect after 4%, 95% or not at all. But eventually, some dump processes get randomly disconnected. This is the sample error I get: [...] DUMP: dumping (Pass IV) [regular files] DUMP: 4.60% done, finished in 1:44 at Fri Aug 18 08:45:04 2006 Received disconnect from 192.168.0.1: 2: Corrupted MAC on input. DUMP: Broken pipe DUMP: The ENTIRE dump is aborted. [...] After several days of googling and pulling my hair out, for some reason I decided to get rid off HTT and SMP... and to my amusement, everything is back to normal. I put back SMP+HTT and the disconnects are back again. So in conclusion, in this case, SMP+HTT causes random problems in ssh. Any thoughts on why this is happening? I really would like to know. What about real SMP machines, are they affected as well? Thanks in advance. Cheers, Mikhail. PS: All servers are 6.1-RELEASE-p3 without go-fast-pc flags or whatever. -- Mikhail Goriachev Webanoide Telephone: +61 (0)3 62252501 Mobile Phone: +61 (0)4 38255158 E-Mail: mikhailg@webanoide.org Web: http://www.webanoide.org PGP Key ID: 0x4E148A3B PGP Key Fingerprint: D96B 7C14 79A5 8824 B99D 9562 F50E 2F5D 4E14 8A3B