From owner-freebsd-questions Fri Apr 14 04:40:52 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id EAA15114 for questions-outgoing; Fri, 14 Apr 1995 04:40:52 -0700 Received: from calweb.calweb.com (calweb.calweb.com [165.90.138.3]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id EAA15108 for ; Fri, 14 Apr 1995 04:40:49 -0700 Received: from web1 by calweb.calweb.com via ESMTP (940816.SGI.8.6.9/940406.SGI.AUTO) id EAA27017; Fri, 14 Apr 1995 04:41:51 -0700 Received: (rdugaue@localhost) by web1 (8.6.9/8.6.9) id EAA17380; Fri, 14 Apr 1995 04:41:50 -0700 Date: Fri, 14 Apr 1995 04:41:50 -0700 (PDT) From: Robert Du Gaue To: questions@FreeBSD.org cc: Robert Du Gaue Subject: Corrupt files and spontaneous reboots Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: questions-owner@FreeBSD.org Precedence: bulk I've been having problems where every now and then for what seems like no reason an executable simply goes 'bad' and starts to core dump and give errors like segmentation fault. It's happened to utils like 'ls', 'groff', 'troff', 'umount', 'mount_nfs', and most recently 'csh'! Taking an original from another machine and comparing, there's usually something like two bytes different. Copying over the executable with a fresh one fixes the problem, but I can never be sure what util will be the next one to go. Not sure if this is related, but I've also been in the office a few times when I've been looking at the console and the system starts syncing disks and rebooting! No error messages or anything, but when this happens it almost *never* comes up on its own. (Pretty tough when using the system as an ISP!) It'll go into single user mode and give me the shell prompt. However, the shell prompt at this point usually doesn't work. A few times I was able to get into a shell by specifying bash as the shell rather then the default (I suspect this is from csh going bad). Once getting a shell I end up reinstalling fresh binaries as the system seems to core dump and go nuts on several different programs after one of these reboots. Installing the bins fixes the problem and i'm at least able to get up and going. So my question is.... Any ideas what is going?????? I have made one change in the kernel which was mentioned here before when someone asked about trying to speed up NFS writes. I've commented out the 'SYNC" param and have recompiled. Is this known to possibly cause problems? I can live with 150-200kbs writes for stability!