I have 4 120-MB SATA harddrives configured in a software RAID5 array
on a SUSE 9.3 Pro machine (ReiserFS filesystem).

There are actually TWO md filesystems configured to use these 4 drives
(md0 and md1).

The first harddrive in the array (/dev/sda) has obviously gone bad. 
This should not be a killer problem (that's what the "R" in "RAID" is
all about).

The second (md1) array will mount in degraded mode (and I can see all
the data).  The first (md0) array will not mount at all.  Any attempts
to mount md0 will cause the shell to hang forever (and I cannot even
'kill -9' the process).

Even if I physically disconnect the faulty drive I cannot mount md0
(though I can still mount md1 in degraded mode).

If I look at /proc/mdstat, I see:
   md1 : active raid5 sdd5[3] sdb5[2] sdc5[1]
         37543488 blocks level 5, 64k chunk, algorithm 2 [4/3] [_UUU]

If I attempt to do this:

   mdadm -A -R /dev/md0

I get a message (after a somewhat long delay) that md0 is added using
3 of 4 drives.  The md0 array then shows up now in /dev/mdstat, but
the array still cannot be mounted (hangs forever). I have also
attempted --force with the same result.

have you tried to mount your drives within a newer linux os , like
opensuse 10.1 or debian? ubuntu live cd should work too -> the md
package on suse 9.3 is quit old (see teh bug list, they provide..)


br,
patrick