On Wed, 8 Sep 1999, David van der Spoel wrote:
> Sep 6 13:10:01 yfs kernel: (skipping faulty sdh1 )
> Sep 6 13:10:01 yfs kernel: (skipping faulty sdc1 )
> i.e., a disk fails, recovery kicks in, and produces a hopefully correct
> array in degraded mode. Then when writing the superblock something goes
> wrong? Does the algorithm that recreates the superblock expect all disks
> to be there?
no, as you can see above, there are already two failed disks at this
point. The SCSI error itself:
Sep 6 13:10:04 yfs kernel: Additional sense indicates Mechanical
positioning error
this is a severe IO error - nothing we can do. The RAID driver shuts the
array down after double-errors and make it unrunnable.
-- mingo
