I'm afraid you'll need a drive partition of at least the same size, since you can't tell for sure how the blocks were allocated on the drive.

Without that, I think you're down to using the --badblocks option and --fix-fixable option. Read the man here if you haven't a machine left to run man on.

http://www.namesys.com/reiserfsck.html

You've past my experiance level at this point.

Regarding the SMART utilities. They should also tell you about accumulated errors. Don't just trust the test. Check the accumulaters after use and watch if they increase. I think there should be a short test and also a long test. Check the accumulators after each test. When you get the machine back up. Webmin has a web-based SMART utility front-end that you can keep an eye on your disk with to catch problems earlier.

Thanks for all the help you've given me Jerre. Anyway the drive was maxtor and I hear they're known to crash at an early age (5 months in my case). It said how you could write over the bad blocks somehow; would you know how to do that?

You're welcome.

First of all, as reiser indicated, the drive should be taking care of the bad blocks itself, the fact that it didn't indicates a bad drive problem.

I think you'll need a combination of the badblocks command to build a list of bad blocks along with the --badblocks and --rebuild-tree option to recover.

When I'm checking for bad blocks on a running system, I usually do something simple minded like fill up the partion using dd while watching the SMART utilites in webmin.

Normally I just trust the drive to work. Lately, since drives are so cheap I never build a linux box without linux software raid.

If you have more than one machine available, experiment with rsync for even more redundancy.