Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Wednesday 2008-06-25 at 08:04 +0200, Per Jessen wrote:

...

If I were you, I'd run some SMART tests instead. Also check the SMART error logs. See smartctl. Hmm, not sure if this works over USB.

I haven't seen any usb HD on which smart is supported :-( They use a very cheap chipset, I suppose.

Hmm, the disk itself should be the same though. According to http://smartmontools.sourceforge.net, there's at least one USB chipset supported with '-d usbcypress' But otherwise it's bad news for USB drives: http://smartmontools.sourceforge.net/faq.html#testinghelp /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Friday 2008-06-27 at 19:25 -0000, Jim Henderson wrote:

...

I haven't seen any usb HD on which smart is supported :-(

They use a very cheap chipset, I suppose.

Yeah, the interface doesn't support it, but the drive's on-board electronics are typically SATA or IDE (these are old enough they're probably IDE) and do support it. Just no interface via USB to turn it on in most cases.

For that reason the external HD boxes I buy now have both usb and sata interfaces, so that I can run smart tests on them. I still haven't tested this feature, I need to add a sata card to my PC. The other thing is to see if I can find boxes with the usb chipset Per said. I have some reading to do. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFIZUcktTMYHG2NR9URAsm5AJ0b7RK2KchYtRhHWxTemzZKmwneZgCgiSM8 GRGuD2yZbuJyv/zUfiRnIPk= =zK2s -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Wed, 25 Jun 2008 10:22:55 +0200, Dave Plater wrote:

If you can see the display of blocks processed and 244,140,625 (1000,000,000,000 / 4096) is the approximate total number of blocks, if you take the time taken so far divided by number of blocks processed times 244140625 you will have the estimated total time. Just subtract time taken so far and you have time to completion.

Now I'm wondering why I didn't think of that. D'oh! And thanks. :-)

If you saved the badblocks list to a file using the -o filename option you can input it to fsck.reiserfs, read man fsck.reiserfs for more info.

Yep, that's my intention.

If smart is disabled on the drives and data was static for a long time you may be able to recover the sectors by doing a few zero write passes but this will take a long time. Its a good idea to run a western digital disk utility on the drives. The biggest cause of bad sectors is data that is static for a long period of time. Regards Dave P

SMART doesn't seem to work over USB, but the data has been relatively static. When you say "zero write passes", is that an option to fsck or badblocks? Is there a Linux-based WD disk utility, or is it a boot diskette (or maybe the utility CD that came with the drive, I'll have to see if I still have that)? Assume I can get that from WD. Jim -- Jim Henderson Please keep on-topic replies on the list so everyone benefits -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Wed, 25 Jun 2008 12:33:53 -0400, Greg Freemyer wrote:

I'm pretty sure he is saying that if you are willing to accept that the data is lost, but you want to put the sectors back in service you can overwrite them with zeros.

Ah, that makes sense, yes. So far only 72 bad blocks, so the data's not completely gone. I'm probably going to try to get ahold of a drive to use to back the data up (one of the problems with drives this size is the difficulty in backing them up).

In deed that should work. The drive itself should have kept track of all the sectors that failed to read. When you write to those sectors, the drive itself should remap the logical sectors to good sectors from a spare supply that it has.

Yeah, one of the suspicions is that SMART is disabled for some reason; I am going to let the badblocks run finish, then pull the drives out of the case and hook them up to an IDE interface directly to try and see if that's the case.

Also, the most recent versions of hdparm have new options related to working with sectors. I don't remember exactly what it can do. The version with 10.3 is not new enough. You need something from the last few months. (8.3 or newer I think?)

You can get the source from http://sourceforge.net/project/showfiles.php?group_id=136732

Thanks, I'll have a look at that as well. Jim -- Jim Henderson Please keep on-topic replies on the list so everyone benefits -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Wed, 25 Jun 2008 21:00:45 +0200, Dave Plater wrote:

Jim Henderson wrote:

...

On Wed, 25 Jun 2008 10:22:55 +0200, Dave Plater wrote:

...

If you can see the display of blocks processed and 244,140,625 (1000,000,000,000 / 4096) is the approximate total number of blocks, if you take the time taken so far divided by number of blocks processed times 244140625 you will have the estimated total time. Just subtract time taken so far and you have time to completion.

Now I'm wondering why I didn't think of that. D'oh! And thanks. :-)

...

If you saved the badblocks list to a file using the -o filename option you can input it to fsck.reiserfs, read man fsck.reiserfs for more info.

Yep, that's my intention.

...

If smart is disabled on the drives and data was static for a long time you may be able to recover the sectors by doing a few zero write passes but this will take a long time. Its a good idea to run a western digital disk utility on the drives. The biggest cause of bad sectors is data that is static for a long period of time. Regards Dave P

SMART doesn't seem to work over USB, but the data has been relatively static. When you say "zero write passes", is that an option to fsck or badblocks?

Is there a Linux-based WD disk utility, or is it a boot diskette (or maybe the utility CD that came with the drive, I'll have to see if I still have that)? Assume I can get that from WD.

Jim

Try badblocks -wt 0 -t 255 -t 0 it will write zeros then ones then zeros again but the best way is a western digital utility zero write over the whole drive. If you run badblocks in read mode once every six months you won't get the problem again. When the magnetic media is at zero it is magnetized and when it is at binary one it is not if it stays at one for too long it gets stuck and needs to be revived. A low level format or zero write does this but it will take a while. If you do a badblocks in read only mode it issues a read verify command to every sector which will prevent this problem. Modern hard drives normally take care of this themselves by replacing bad sectors from a pool but only if smart is enabled. This must be a problem with usb hard drives. If you take the drives out of the casing and put them into a computer and enable smart you might find they are still under warranty if you visit WDs website. Regards Dave P

Good info, thanks. So far I've not lost that much data (only 72 blocks so far, but the read-only run has been going for nearly 5 days now), so I'm not sure yet that I want to blank the drive (esp. since I've nowhere to back the good data up to as of yet - but I probably will in the next week or so). Jim -- Jim Henderson Please keep on-topic replies on the list so everyone benefits -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Fri, 27 Jun 2008 15:37:13 -0400, Larry Stotler wrote:

To get some good drive utils, download the Ultimate Boot CD:

http://www.ultimatebootcd.com/

IIRC, it includes the WD utils.

Cool, thanks, Larry - appreciate the pointer. Jim -- Jim Henderson Please keep on-topic replies on the list so everyone benefits -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Sat, Jun 28, 2008 at 3:32 PM, Jim Henderson wrote:

What I'm thinking is that if I can get a list of the files that are affected, nuke them, and then refresh only the affected blocks, that may well save me a bunch of time and money in backing the drive up (which means buying another drive = the cost factor). The errors reported in the log are data phase errors (at least for where I've tried to write), so it *seems* reasonable that I could probably do this without wiping the whole drive and starting over.

Probably not a good idea. I had a drive go bad(on the reiserfs fs table no less). Even after running bad blocks, I still don't trust the drive. I would be very leery of it. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Saturday 2008-06-28 at 19:32 -0000, Jim Henderson wrote:

On Tue, 24 Jun 2008 23:02:19 +0000, Jim Henderson wrote:

...

Got a drive that's starting to show its age (sadly, at only about 2 years old!). It's a Maxtor OneTouch drive, which actually contains 2 500 GB drives and presents via USB as a single storage device.

Two follow-up questions:

1. Is there any way with reiserfs to find out which files are affected by the results of a badblocks test? ie, given a list of bad blocks, can I find out the filenames of any files that are going to be unreadable as a result?

Maybe, but I don't know it. An alternative is to read every file till it fails, then delete and replace the file.

That'd give me an idea of what I need to find backup copies of. Most of the data is replaceable, just a pain to have to do that.

2. Suggestion was made that I could maybe re-zero the drive to revive things. Could I just zero those blocks in the same way, and if so, how?

Better copy everything somewhere else, and then overwrite the entire drive, and reformat it. The overwriting on normal disks triggers the bad sector remapping, but I don't know about usb enclosures. I suppose it happens, the only thing is that you can't interrogate the chipset about it. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFIZ2JftTMYHG2NR9URAhkuAJ9qwDcZQA4zaIp7sKm7NI2cwWAYpwCcCF7h SN08JN7i/tAsnCzXEPgHEoU= =fKHj -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Sunday 2008-06-29 at 16:34 -0500, Rajko M. wrote:

The reason was that 10.3 that was asking too much from old drive and long data transfer overheated electronics, so it started to write garbage on a disk. I didn't found the way to tell 10.3 to slow down and changing BIOS settings (disable UDMA) wasn't regarded. So I tried to rescue files and move them back to original drive. This made situation worse. Drive started clicking and disappeared from the system with majority of files on it.

Perhaps with hdparm: "-d 0" should disable the dma. Or tinkering about you might convince it to use a lower dma setting, I think. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFIaAiMtTMYHG2NR9URAsjiAJ9LMOzUcmSiJ2qt8VGGuwbrvMTc2ACdGuVy vkh2e/hNI/vwqzs/cyHQWlY= =/pJt -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Sun, 29 Jun 2008 21:35:26 -0500, Rajko M. wrote:

It can't be worse than it is

That certainly sums it up... Jim -- Jim Henderson Please keep on-topic replies on the list so everyone benefits -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org