I just installed raid 1 feature for the small disks (2 x 8 Gb)
on my production server (for the large ones, 4 x 50 Gb, i'll wait a bit
since reloading the datas in case of failure would require to feed many
many CDs whereas I have an additional disk to disk backup for the small
disks)
When booting 2.0.37 kernel, the boot process stopped and asked for root
password in order to get to a shell instead of normal boot.
This is a bad behaviour since no RAID device appears in /etc/fstab so
a problem with the RAID drives should not prevent normal boot: the
result for me was very bad since I use 'vnc' to remotely configure the
server, so stopping the normal boot process made me loose the server
control (the server is 100 miles away from me).
Switching to a shell instead of normal boot can be a reasonable
behaviour when one / partition is damaged, but not when there is
a problem (moreover in this case it was a virtual one) on a partition
which is not mounted during the boot process.
So I had somebody insert a floppy in order to boot through NFS root,
I renamed /etc/raidtab to /etc/raidtab0 and rebooted again from the hard
disk.
The boot process went fine, but then 'raidstart -a' complained something
like 'bad argument' or 'bad device' (sorry i don't remember the exact
message)
Lastly I switched to 2.2.12-pre4 kernel and everything ran just fine.
Regards,
Hubert Tonneau
