I was doing a kernel recompile today on my Gentoo fileserver, and after it had built, I set about re-emerging video and sound card drivers.. I came back a few minutes later expecting it to have finished, but instead I had a "segmentation fault" error waiting for me.

Things looked to be quite screwed up after that. It couldn't switch runlevels, and so I had to reboot it by hand. And then, my fears came true.. It looked like my root filesystem had corrupted itself.

It was formatted with reiserfs, so I booted my livecd and ran fsck, which couldn't do anything except rebuild the entire tree. A few minutes later, that resulted in the partition containing numerous unhelpfully numerically-named directories with all my files in. In no particular structure at all though, it seemed.

Luckily it was just that drive that was corrupted, and I didn't lose my 40GB+ of music and video. The machine's currently being set up with a fresh new install (and separate /usr and /home partitions this time), but this really makes me worry.

Is this most likely the fault of a dodgy hard-drive (the one in question is a 13GB Quantum drive, a few years old), or a bug in reiserfs, or what?

I'm quite worried about losing my data here, and currently I don't have any efficient ways to back up over 40GB of data.

Could anybody suggest what might have gone wrong here, and how I might prevent it happening again? I used to have windows on the same machine for a few years and I can recall one time when it found bad sectors on the hard drive, so that's a possibility. I'd only had Gentoo installed on the machine for a few weeks before this happened.

What linux tools are there that I can use to locate bad sectors on my hard drive, and are there any other things I should be doing to better protect my data?

Thanks in advance,

If its a quatum fireball, they had a whole string of those that went bad, and the 13 gig was one of them, I have personally replaced over 100 + of these drives. What happens to them is a bearing or something goes bad and the drives really start clicking then you can only hope to have enought time to get the data off before the drive is dead.

I hope this is not the problem yours is starting into but I have seen this many many times with the 13 gig quantum fireball's

Let me guess, Compaq?
At any rate you couldn't give me a Quantum or a Maxtor.

Maybe they need to be confined to the grave then. Heh, I didn't think it could have been a bug in reiserfs: that's just unthinkable .

Cooner: Not a Compaq then, heh. But what's wrong with Maxtor? I'm running a RAID 0 configuration using two 40GB maxtor drives and I've never seen anything less than excellent performance from it..

Thanks people,

Still.. are there any linux tools for proper hard disk analysis?

http://www.maxtor.com/en/support/downloads/powermax.htm

That is the utility you need to test the drive correctly, maxtor bought out quantum so yes this is the utility you need, you will need a windows machine to make the disk, but once it is made you can use it on your machine linux or no linux. If you don't have access to a windows machine, I could mail you the disk if you want.

Oh and if you really want to test the drive at the drive level I think it is always better to do it with no os loaded, that way it eliminated the possiblity of a os problem or corrupt filesystem, as with this it will test the drive itsself.

quatum fireball harddrives have a lot of problems don't they , my sister has a 50GB fireball hardrive brand , I split it with windows and linux , I set it up so she could use linux for her net access since she always seems to miss it up in windows , anyway , the hardrive is slow to read in linux in the kernel bootup progress it says dma reset message , it takes linux 2 times before it gets passed the message in bootup, I put Suse 8.2 Pro on her Sony computer , I only thing about linux and & Sony computer problem is their no fan controll program in linux , as far as I can tell that makes the fan less sounds thing, windows has a driver for Sony Desktop computer that puts the fan sound down or something like that it controls the sony fans in windows only. Linux seems to run ok even thou theirs no fan controler thing for linux. MY older Pentium 3 Sony computer had no fan control so linux runs great on my one anyway.

Ahh, that sounds just like what I need! Many thanks, I'll give it a go..

Do you know if they're replacing or (my prefered) refunding fireballs gone bad? I've got a sixty gig Fireball Plus AS (coulnd't seem to find it on the site). It was about a year and a half old.

Those quantum fireballs, are those the ones that catch on fire when they hit 16000 rpms or was that the maxtor?

All the drives I replaced were out of warrenty with quantum, some of them got replaced under the gateway warrenty, but unless your computer or drive is still under warrenty you are just out of luck :-(

Oh and I should hope the drives don't end up spinning up to 16000 rpm because they are only at most 7200 rpm drives :-) so if they get up to 16000 rpm they have every right to explode into a fireball (sorry couldn't resist, you know blow up, fireball, quantum fireball :-) Don't worry I will be here all week, thank you, thank you very much)

The nasty utility bums out with only a "Cannot load DOS!" error message, which isn't particularly helpful to me, ugh.

Disregard my previous post.. got it to check both hard disks.

The first disk (Quantum Fireball CX13.6A) passed all the tests, much to my annoyance, and I'm currently running tests on the second disk (Quantum Fireball CX13.0A) right now for safety. I've never encountered problems on the second disk, only the first. However, the fact that it has passed all the tests is a little worrying - I'll perform some of them again just to make sure.

Might as well buy another hdd - they're so cheap now anyway .

You might want to run it 2-3 times in a row, the problem might be one that shows up after the drive warms up.