On Thu, 26 Oct 2000 [email protected] wrote: > on Thu, Oct 26, 2000 at 11:13:44PM +0200, Shaul Karl ([EMAIL PROTECTED]) > wrote: >>> A week or two back it started misbehaving and I asked a few of you guys >>> about >>> the place about what "Unknown vector XXX in CPU#0" and "hda interrupt lost" >>> meant.. >>> >>> Some people said that it sounds like the hard disk is on the way out. What I >>> want to know is, how do I test an ext2fs formatted hard disk more >>> intensively >>> than at just a filesystem level?
>>> If it is indeed the harddisk that has died, does anyone have any >>> good-condition 1GB-8GB IDE drives? I don't think this old 486'll handle over >>> 8GB, that and I'm not too crash hot on using some of that 'patch my bios on >>> boot' master boot sector voodoo evil :) IIRC, you are located in Australia. Not sure you'd want to pay shipping for my old drive. ;) >> e2fsck > > I'll have to disagree. Agreed. [Side note: one of my pet peeves is people posting incorrect information to the list. Of course, more annoying is when they post a correct response but to the wrong question. Not sure which this one was, but....] > Many professional system administrators strongly recommend replacing > hard drives at the first sign of failure. This may be overkill, but > given the relative values of hardware to data contained, it probably > makes a lot of sense. Think of it this way: you value your time at $100/hour, and a user's time at $1/hour. Say you have 100 users, and there's a hard drive failure on the partition with their home directories. You now face minimum 1 day downtime while you replace the drive and restore from backups. Your users lose 1 day's previous work, plus can't be productive for another day. Assuming they actually work 6 hours/day, that's $1200 right there. Factor in your time of 12 hours to get it back online, and it's another $1200. Would have been simpler to replace it earlier, costing your time only (and less of it). Of course, if it's a home system and you're the only user, most of this doesn't apply. Just keep regular backups and watch your system carefully. > Other recommendations on hardware testing appreciated. I agree badblocks is probably the best, but you could also try bonnie and bonnie++. If you want to check the health of sectors already occupied by files, I suppose a dd if=/dev/hda of=/dev/null wouldn't hurt. Damian Menscher -- --==## Grad. student & Sys. Admin. @ U. Illinois at Urbana-Champaign ##==-- --==## <[EMAIL PROTECTED]> www.uiuc.edu/~menscher/ Ofc:(217)333-0038 ##==-- --==## Physics Dept, 1110 W Green, Urbana IL 61801 Fax:(217)333-9819 ##==--
