[Wylug-help] RAID 10 Array - possible solution, please confirm/deny

[Dave Fisher]

Since I last posted, I've made a bit of a breakthrough.

If I try:

  $ mdadm --assemble /dev/md1 /dev/sdb4 /dev/sdc4 /dev/sdd4 /dev/sde4

I get:

  mdadm: /dev/md1 assembled from 3 drives and 1 spare - need all 4 to start it (use --run to insist).

This is obviously promising, and reminds me of something I read somewhere about
bad metadata on the first drive's superblock potentially screwing up mdadm.

Before I try using --run, I'd like a bit of re-assurance about it's effects,
because the man page description is uncomfortably terse.

I guess that using --run will cause mdadm to use parity data from the other
three drives to rebuild the fourth from the spare, overwriting blocks on the
spare.

I'd be happy to try that, if:

   1. There was a high degree of certainty that the rebuild would be completely
      successful, or;

   2. I could be certain that mdadm has now correctly identified the spare.

I don't want to go from the current situation (in which I seem to have 3 sound
partitions + 1 bad + 1 spare) to a situation in which I end up with 2 good
partitions + 2 bad + 1 spare.

My current guess is that sda4 is the bad partition (cos mdadm plays nicer
without it) and that sde4 is the spare (which is what I thought I remembered
from the original build).

Any thoughts?

Dave