Followup on the Dual HD + RAID1 challenge


Seems like either I didn't express my question very well, or it
turned
out to be a tough one.

Anyway, an update on the situation:

I could manage to make it work on a test machine configured with
/dev/md0 mounted on / and /dev/md1 mounted on /home

[recall that the goal is to take a machine that has two hard disks,
all
configured as RAID-1, and disable some of the RAID-1 partitions and
make them behave as independent ones --- all this without physical
access to the machine; just remote root access through SSH]

I boot the machine, login as root, unmount /home (umount /dev/md1),
then change in the /etc/fstab, the line that reads /home /dev/md1,
I
change it to /home/sda2 --- sda2 being one of the components of the
RAID-1 device md1). I then delete /dev/md1, but still after a reboot
the system keeps seeing a /dev/md1, and then the mount for /dev/sda2
fails (since it is busy, being part of an array).

I then tried creating a new filesystem on /dev/sd2 (mkfs -t ext3 /dev/
sda2)
and on /dev/sdb2 (the other half of the RAID-1 device md1), and the
problem went away --- my machine now boots, and I have /dev/md0
mounted to the root directory, and then /dev/sda2 mounted to /home.

My question at this point:

Is the above a catastrophe waiting to happen? I mean, did I do
something
wrong that worked by chance/coincidence but that could as well stop
working
any time soon given some particular conditions?

Any comments will be appreciated.

Thanks,

Carlos
--

.

Relevant Pages