From: Michael M. <mi...@np...> - 2004-12-21 05:17:56
|
Hi Bruno, > > What I've done after our discussions from yesterday was to use your commands > > below as: > > > > # mdadm /dev/md0 -f /dev/hda1 -r /dev/hda1 -a /dev/hda1 > > > > for each md device, which successfully went through and re-synced the mirrors. > > > > I then booted a rescue cd and: > > > > e2fsck -c /dev/hda1 to 8 (except extended and swap) > > > > No errors were found. > > It would be pretty unlikely to hit something there, since that > doesn't check the data in files. > > > I did this last night with "e2fsck -c" as discussed above, but it didn't find > > anything wrong with the drive. Should I be using the "badblocks" tool for > > this? > > That should run badblocks, but you might want to run it yourself > since you have more flexibility in the test. If the disks aren't > that big letting badblocks repeatedly read and write the disk might > be a good idea. With bigger disks this can take a long time. The disk with issues is only a 40Gb disk, so it shouldn't be a drama to hit this with multiple passes of badblocks tonight. > There might be something odd about that model of drive. Once a successful > read has been made from a blocking that has a pending reallocation, > it should be reallocated. It's a Seagate ST340016A 40Gb drive, and I've software mirrored it with an ST340014A 40Gb drive. They are essentially the same although the ST340016A has an 8ms access time while the ST340014A 40Gb drive has a 5ms access time. Tonight, I'll boot into rescue mode and run the following: # badblocks -c 32 -p 20 -n -s /dev/hda If you see any issues here please let me know. Michael. |