Date: Wed, 25 Feb 1998 14:24:43 -0600 (CST) From: Jeff Lynch <jeff@mercury.jorsm.com> To: freebsd-isp@FreeBSD.ORG Subject: Re: BSD crashes under load Message-ID: <Pine.BSF.3.95q.980225134527.19982C-100000@mercury.jorsm.com> In-Reply-To: <Pine.NEB.3.96.980220141246.4419A-100000@ns1.netcorps.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 25 Feb 1998, Satya Palani wrote: > Recently, we've been having a large number of crashes on our FreeBSD 2.2.x > web servers; unpleasant crashes, where someone has to come in and fix it > at the console. Specifically, it's happening on our web-hosting machines > (running either 2.2.1 or 2.2.5), each of which are hosting several hundred > customers. Same here on 2.2.1-RELEASE. Twice in the last month. Slightly different symptoms. > > The problem seems to be disk related. There is no panic message in the > log (or any message, for that matter); the machines simply spontaneously > reboot, fail the filesystem check, and drop into single-user mode. At In contrast, we do get a kernel panic a message to the effect that the system will reboot in 15 seconds (which it does not, physical reset is required). But we do not have to go single user, although we should, but our filesystem corruption has not been as bad. > this point, running fsck brings up a long list of duplicate inodes/files > in the /usr slice. I would initially consider this to be problem with the Upon reboot, fsck spits out a list of dups and no other corruption serious enough to require manual response to fsck and everything comes back up fine. > drive; however, replacing the drive on one server didn't fix it, and > another machine that was just added a week ago is starting to display this > behavior as well. This sort of thing should not be happening on a > brand-new drive. > > So, what it looks like to me: different processes are trying to write to > the same disk sector and are killing the machine. Since we have a lot of > customers ftp'ing their sites to the servers, there is a *lot* of disk > activity going on, and I'm wondering if it's too much for FreeBSD to > handle... Does the OS get confused if more than a few people are > transfering data to or from it? I'm wondering if it's the latest Pentium bug causing the system to panic. We allow only very little shell/cgi activity on this server so we are looking for this but it's not a highly likely cause. Probably more likely to be marginally failing hardware somewhere on the system. Interesting this just started happening to both of us, so I decided to throw it out as another data point. > > Has anyone else experienced anything like this? We're using Adaptec 2940 > SCSI controllers and Quantum Atlas II drives, so I don't think hardware > quality is an issue... Same here, 2940UW with Seagate drives. I wish I could rely on a diagnostic program to actually check the RAM sticks. ========================================================================= Jeffrey A. Lynch, President JORSM Internet email: jeff@jorsm.com Northwest Indiana's Full-Service Provider Voice: (219)322-2180 927 Sheffield Avenue, Dyer, IN 46311 Autoresponse: info@jorsm.com http://www.jorsm.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95q.980225134527.19982C-100000>