Hard disk light stays on. No processes, no apps running. reboot=corruption. wtf?

I found one similar thread, but no conclusion or followup..

After a while, usually once a week, I notice the hard drive light is on. It's on steady, but there doesn't appear to be any activity going on.
The system seems a tiny bit sluggish while this is happening.
About a month ago, it did this, and after about 15 or 20 mins, the light went off.

I closed all applications, checked ksysguard for weird processes, and didn't really notice anything odd. The northgate temperature is low, indicating no IO activity. I reset both HD's with hdparm -w. I logged out, went to a console and switched to telinit level 1. All processes were really killed now, yet it was still on. I ran TOP and still don't see anything suspicious.
Finally I got fed up, and typed 'reboot'. Sheesh, it took me about 4 fsck's and a few reboots to get the filesystem repaired successfully. Finally I rebooted, got back into what I was doing (which incidentally isn't very much -- browsing, updating slackpkg, typing a text file) and now, here the damn thing is on again.
This time I won't reboot, but I'd like to figure out wth is going on here.

I reset my BIOS to factory defaults before this reboot. The main HD (ide0,a) is a brand new Seagate 320 GB, and hdc (ide 1,a) is an old WD 14GB.. The whole system generally works flawlessly. It's an Intel P4 1.8 Ghz 845PE, about 3 yrs old. Runs Slackware 11, 2.6.20

I'm annoyed, and worried. If anyone has any idea what to check or where to turn next, I'd be very happy for any advice or suggestions..


It's still on....

UPDATE: I posted this initially at 23:35 (server-time) and just now (00:38 server-time) I ran 'lshw' for the heck of it, and as soon as the results scrolled up the terminal, the light went out .. Coincidence? I dunno.. A few minutes ago I killed all the KDE background services; that didn't change anything. But now, after lshw, the light is out.
Must remember that for next time :/

I had the same drive... new...same problem...and it died got a western digital replacement. might want to download the hard drive utilities from seagate and check it out.

Stupid Seagate website has issues -- maybe they are using their own drives
I can't download anything from it, it keeps giving me a "404 - Cannot find design notes 546533677545.374659912746502374" type of message.

Thanks for this suggestion Hern_28, though it isn't en encouraging prospect huh? LOL, defective brand-new equipment is never pleasing. I will keep trying the website to see if I can't get the thing to download. By chance, have you got (a copy of) the DOS version of the Seagate tool you are referring to?

Thanks

UPDATE - downloading a bootable .ISO of a Seagate diagnostic tool from http://www.computerpi.com/seagate.htm.

Fingers crossed folks.. I hope this doesn't make things worse, as in mess up my installation..

Finding the problem early is a good thing. i caught mine soon enough to just copy linux onto the new one so i didn't loose anything. what my drive was doing was failing an identification test which set the computer into a loop trying to id the drive. (guess is was the mb trying to fix the problem.) let me know how it works out . might also want to double check the jumper settings on the drive(s) to ensure they are set properly (remember don't use cable select).

According to Google, drives less than six months old are just as unreliable as drives that are three or more years old, regardless of manufacturer. I guess it's just cheaper to replace the bad ones than burn-in test them for a week.

Anyway, I know you're convinced it has something to do with the new drive, but maybe the problem has something to do with moving the WD to your secondary controller. Is there a CD/DVD drive on the same controller operating in PIO mode maybe?

That light flashes from controller activity actually, not hard disk activity as you might assume from the name of the jumper (HDD LED).

Hi Hern & Crito - thanks for the support.

The jumper settings are definitely correct; been in there a few dozen times and NO cable select. Just Masters and slaves, and in the right orientation along the IDE cables. I will let you guys know what becomes of this. I can't afford a new drive though, so if anything shall become of this as far as replacing this one, it'll hafta be a 'defective/exchange' scenario.

Crito - Agreed, though this one is now a couple months old, it seemed fine at first. And interesting about what Hern_28 said about maybe the mb trying to ID the device, because the shit stopped the very second when I ran lshw (I actually thought lshw would hang).

Incidentally I just ran lshw as I write this post, and it hung on reading the IDE. Still hanging. Usually it works fine. CTRL-C stops it. #!$@%#^$

Crito, the WD drive has been on controller#2 from day one since I got this drive, and AFAIK all 4 devices are operating in DMA/UDMA modes:

0,0 Seagate ST3320620A 320 GB *UDMA5
0,1 ATAPI CD-R/RW             *UDMA2
1,0 WD 14gb drive             *UDMA4
1,1 ATAPI DVDROM              *UDMA4

Any further input is gratefully appreciated. After I have lunch here very shortly, I'll burn this bootable Seagate tool and see whats up with it. Will let you'se know.

To remind yourselves why you love Linux/Firefox, plop yourself down at a WinXP/IE machine for a mere 5 minutes, and I swear, you'll be fit to be tied!! This is sheer ridiculosity!!

Well, on topic: The Seagate HD failed the 'Quick Self-Test' issued by Seatools Desktop, and is now plodding its way through the 'Full Test' which looks like it will take 2 hours, give or take.
I called the awesome little computer store where I got the HD, and with little ado, the guy told me they have another one in stock, and to bring it in for a no-questions-asked exchange Yay! Hopefully I'll do that this weekend.
I do like the drive, and the mfgr, so I'm not worried about a series of crappy Seagates coming my way -- even top of the line stuff is borked sometimes .

Finally, if one of you guys/gals would be so kind: Of the myriad ways to backup a system, whats the SIMPLEST way to clone my entire Slackware installation onto a CD, with the intention of cloning it right back identical to how it is, on the new drive? I'm thinking something like 'dd partition --> file.iso --> cdrecord it /.../ dd the CD --> the new partition' then boot from my Slack DVD to fix up my LILO/MBR.
How's that sound?

OK, well I HAVE TO GET OFF THIS WinXP MACHINE before I try to inject myself with drain-cleaner to end the pain! Thanks folks, see ya soon -- from Linux

Sasha

Was just making sure have had interesting problems with hard drives simply because i forgot to change a drive from master to master with slave present . Took me about 2 hrs to decide I was an idiot lol.

Could use tar or create another partition and use rsync to keep a running backup system and use chmod to automatically update it(still figuring this one out, currently using the tar method).

Originally Posted by Hern_28 Was just making sure have had interesting problems with hard drives simply because i forgot to change a drive from master to master with slave present . Took me about 2 hrs to decide I was an idiot lol.

LOL, been there a few times too
OK, looking at tarring/gzipping the entire partition as one giant archive, to the other drive (as obviously another partition is no good - the drive's going back )

Now, to try (partly) answering my next question on my own: When I put the new drive in, and make my partitions, here's my plan:

1 - Install a basic Slack from the CD/DVD, to simplify setting up fstab & MBR initially.
2 - Boot a LiveCD, and extract my backup archive into the new root partition.
3 - Reboot using the Slackware CD/DVD and rerun LILO.
4 - All done?? Would this do it? Sounds too easy.

On second thought.. I don't even need step #1 do I ?
P.S. - rsync sounds great.. Something else to get aquainted with!

No don't need step one. Need to edit the fstab to prevent errors on reboot and seem to remember needing to fix the swap configuration. Don't know if I just been getting lucky or forgot a few steps (haven't irrecoverably crashed Linux in about a month now) . Would recommend getting the new system up before sending that other drive back.

Well, I don't plan on being without the drive for more than a few hours. So, I don't plan on having the system running while the drive is out of the machine, ie: I'm not gonna run the machine off the other drive. (Though that's a neat idea.. Hmm..)
Yes, you're correct, and I have been conveniently lucky too -- As long as the fstab resembles what it should re: the new partitions laid out on the new drive, it's no big deal. Basically, I figure as long as the swap and root partitions are correctly identified, the rest can wait till I get booted to fix them.
I'm just bzipping my home partition now, and will bzip the rest of the OS next, and save them both to the spare HD.
Then, after partitioning the NEW HD, I'll boot Slax or something, and extract the archives to the new HD, and then use the Slack CD to boot the newly filled root partition. Then fix fstab and repair LILO, and that should be it
Woohoo! I hate waiting, but.. Well, this will all happen this weekend I hope, so -- soon enough!
Thanks for the continued support Hern_28

Just returning the favor

If the other drive is large enough to support the install and the tar-ed file why not I would be curious as to how difficult it would be to get it to work. its the same system so maybe just editing fstab will work. have never tried it though but would be interested in whether or not it would work.