On Mon, Jan 07, 2008 at 03:58:47PM +0100, Ivan Savcic wrote: > On Jan 7, 2008 3:37 PM, Douglas A. Tutty <[EMAIL PROTECTED]> wrote: > > > Even if the filesystem type didn't need to be fsck'ed, for a damaged > > drive I wanted to try to reuse, I wouldn't put real data on it until I > > had exercised it for 24 hrs straight anyway. > > I agree, stress it as much as you can, ie. create partitions before > and after the damaged part of the drive, create a 512 MB file from > random data, then fill up the drive with that file (do the copying of > the file in parallel). Then, again in parallel, do md5 sums of those > files. Doing copying in parallel and then md5 summing in parallel will > stress the drive with seeking. If it completes successfully, I think > this drive would be "ok" to use for a year or so if you're lucky. >
With bad-block remapping, you don't really know where the "damaged part of the drive" is. I would go with one whole partition and if after fsck -c -c there were still errors showing up in syslog, I'd ditch the drive (after sanitizing it). The fsck -c -c should make a manual md5 check redundant. You can also use wipe as a simple command line that will overwrite the drive's surface many times as a way of giving the drive a bit of a workout. It doesn't do random seeks though. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
