I posted this this morning, and it didn't show up yet. So, I am thinking
maybe I made a mistake in posting it. (Sorry if it shows up twice.)
This is my first visit to this newsgroup, so please forgive me if this
question has already been asked and answered. I have a RAID5 array on an
H70, and recently (per users request and IBM's recommendation) had a new
disk installed in an empty slot in the RAID5 array. I actually think I have
two new disks there -- both 18 gig. The old disks that were originally
"raided" are all 9 gig, but IBM told me it was okay to add 18 gig disks to
I opened a PMR with IBM support on this, asking them what the steps are for
adding a new disk to an existing RAID5 array. Their initial answer was that
this can not be done. Then they followed up by saying that I can REPLACE an
existing disk with a new one, and gave me a procedure. The procedure
seemed to imply that I CAN add a new disk though.
Basically the procedure was to:
1. Backup all filesystems.
2. Remove all filesystems, volumes, and volume groups.
3. Use smit to define the new disk as a "candidate"
4. Use smit to make a "pdisk" out of the candidate (not sure -- am writing
this from memory)
5. Add the new pdisk to the array.
6. Re-create the volume groups, logical volumes, and filesystems.
7. Restore the filesystems.
With about 50 gig of data being involved, this is a weekend job. I need to
know that this procedure is correct, and I need more specific instructions.
For example, I know that the "hot-swappable spare" must be as big as the
biggest disk, so if I do have two 18 gig disks, I need to make one of these
Another question I have is about backing up the data. We use ADSM (now TSM)
as our main backup system, but I believe that making a full backup to some
other system disk would be a wise precaution. Should I do an rdist, or a
smit backup, or ??? I don't think tar would be good, since I heard that tar
does not preserve the acl's on the files it backs up..
Thanks for reading all this, and for any help you can offer!
- Gabe Strasser