I'm looking for a solution for storing backups on a 750GB drive that has had some reliability issues. I'll have backups elsewhere also, but this will be a convenient place to store backups on a more frequent basis. I know the dangers involved with what I'm asking, but this is also a bit of a learning experience.

What recommendations or solutions are there for storing data long term on a drive that might have issues? Something like a drive-wide PAR solution seems the most feasible and effective. An optimal solution would be windows based, but I have no problem install Ubuntu or something else on the drive.

Any help or advice you can give would be greatly appreciated!

Thank you for your time.

Caesar Kabalan

Some unnecessary background information:

An older Seagate 750GB Drive I've had for a number of years has been giving me some slight issues lately. Several months ago Hard Disk Sentinel gave me some information that problems during disk spin up have been happening. I immediately backed up all the data from the drive. I recently bought a new 2TB to replace another drive, which will be replacing my Seagate. The drive has been functioning normally within the OS for several months. The only hint (other than Hard Disk Sentinel) that anything is wrong have been 20sec hangs when booting while it searches for SATA drives. This happens on 70% of boots.

The drive doesn't seem to be experiencing any bad sectors. This makes me think this whole exercise may be futile, as I'll boot one day and the drive just won't spin up. No amount of error checking can recover from that. I do have another older 500GB drive that IS experiencing an odd clicking noise that I might want to try something like this on also. Again that's likely an issue with the head and might just fail catastrophically one day.


So, as long as you think of this drive as a local cache for your backup system, then, probably no problem.

I would store, both on the dodgy drive, and on a good drive, a sha256 hash of each file you back up. That way when you copy a file back, you can sha256 sum the copy and if the hashes compare you got the file you expected back. You can get a windows program to do sha256 hashes at http://md5deep.sourceforge.net/.

Otherwise, if you can't copy the file back (ie, the drive does tell the OS you have an error) OR if the hashes differ then you know that at least that file is bad.

You could run a task every week/month/whatever over the drive recalculating the sha256 hashes and if they start differing know that the drive might not be worth it anymore.

Amusingly, the base error rate of a modern sata drive (WD30EZRX, 1e14) means that if you read the drive 100 times completely you will likely get an unrecoverable error. This has implications for how we do think about even shorter term storage.

  • I'm familiar with MD5 sums. I assume SHA256 isn't much different conceptually. So your solution isn't necessarily to repair data but rather to assume it is functioning at 100% and keep checking on the data. When you start getting an unreasonable number of errors, scrap it and buy a new one? – Caesar Kabalan Sep 16 '11 at 9:13
  • Yep. Actually I'd buy a new drive now since I like to have more trust in my disk drives... If you want to repair the data then store more than one copy of the file, and, if one is bad, try reading the next. Whether that works well or not depends on why the disk seems to be going bad. – Bruce ONeel Sep 16 '11 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.