fsck or badblocks?

[Firewing1]

Hey,
every once in a while I get this in dmesg after the computer freezes for 10 s and makes a funny HD noise:
hdb: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hdb: dma_intr: error=0x40 { UncorrectableError }, LBAsect=9671186, sector=9671176
end_request: I/O error, dev hdb, sector 9671176
hdb: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hdb: dma_intr: error=0x40 { UncorrectableError }, LBAsect=9671186, sector=9671184
end_request: I/O error, dev hdb, sector 9671184
hdb: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hdb: dma_intr: error=0x40 { UncorrectableError }, LBAsect=9671186, sector=9671184
I've had this before, and a reformat fixes it. I did fsck -Cf /dev/hdb3 (my root) and everything's OK. How do I fix the HD error?
Firewing1

[Furball]

Hello Firewing,
does badblocks show any errors ?
You could try 'badblocks -s -n /dev/hdb' to do a non-destructive read-only test. You can scan certain blocks only as well but I don't know how to translate the sectors from your error message into the right block numbers

[Firewing1]

It's happened a couple times; every time I would fsck and bacblocks and neither would show any results.
It's just annoying cause the only way (i know of) is re-format, and im not doing that.
Firewing1

[Furball]

That's strange. I had some ibm harddisk some time ago with the same "symptoms". If fsck and badblocks don't work you can try the manufacturer's utility. Most of them have a special utility downloadable on their website which can test and stress-test your drive. They work only for the drives of the given manufacturer but they can be very powerful like they can read some internal stored error codes which fsck just cannot do. Might be a bit difficult to find on the website however (ibm, seagate and maxtor do have them, I know).

PS: My harddisk died after 3 weeks beginning with making these noises from a head crash

[Firewing1]

lol.. thanks... although it is a faily old HD it's a quantum Fireball AS 20 GB but it's been doing this for 2 years, as i said, every once and a while... so i'm not worried about crash. Plus, check out my sig, and take a look at my Backups program. It's what I use. I had a full sys crash once, had to reformat, and I cloned my system -- ended up with EXACT same packages, configs, documents in under 4 hours. I love my backups. Even better, it's done for me at 1am every day onto my USB HD; The hour is configurable by the user.
Firewing1

[Firewing1]

OK this is really annoying. I found the glitch; short reads in my ext3 fs. I rewrote them, now it's ok. But, the system's been acting really slowly ever since; XFS takes 10 seconds to start and it normally takes 1 or 2 MAX. Same for crond, cups, etc. I tried dmesg, nothing. The system will also freeze will accessing HD on occasion, but no errors show. Should i reformat completely?
Firewing1

[Tashiro]

Hi Firewing,

Make backups of you data, if you haven't already. Sorry I don't have any better suggestion.

Tashiro

[Firewing1]

I do, every night with my FWBackups program (See my sig)
I'm just wondering -- is this a sign that my drive is failing or is this normal?
Thanks
Firewing1

[Firewing1]

EDIT: OK, my real question is -- Will reformatting fix this. I've been getting these errors for over 2 years, and it *seems* every time I reformat they go away, and slowly come back. I just did an fsck, it shows nothing. Yet I still get these HDB errors.
Firewing1

[Firewing1]

This is annoying. I reformatted my /boot (/dev/hdb1 was were the errors where coming from. Still errors. So i guess the surface is scratched !?
Anyway, when I boot HAL takes no joke 15 seconds to start, and it posts errors about block 104385 / 104389 in /dev/hdb1. Those are on the badblocks list. Why does it keeps accessing them?!?
Firewing1