Hi,
only two hints:

to copy the partition table from one to an other disk:
prtvtoc /dev/rdsk/c?t?d?s2 | fmthard -s - /dev/rdsk/c?t?d?s2


make the disks bootable:
/usr/sbin/installboot /usr/lib/fs/ufs/bootblk /dev/rdsk/c?t?d?s?

Br Joerg

Hi,

Is the replacement of the two 72GB drives w/ two 146GB drives just
because you can get the drives quicker for replacement?

Because you dont use the additional added space from the step to 146GB drives, I can write a little how-to of doing that if you want..

In short :

-) Break the 2x72GB mirror by removing the second disk (c0t1d0)
-) Remove the drive and insert a 146GB drive
-) Copy (for now) the VTOC from disk1 (72GB) to disk2 (146GB) OR
create the needed existing partitions by hand and create a big (or two/three..) slice/partition to hold the additional ~65GB space
-) Create metadb replicas on the 146GB disk
-) Start mirror sync
-) When mirrored, break the mirror and remove the first (72GB) disk, V240 now runs from the 146GB disk in slot 2
-) Insert the other 146GB disk and copy the VTOC from disk2 like stated from jörg via prtvtoc , I always add the additional '-sh' arg to prtvtoc
prtvtoc -sh | fmthard -s -
-sh removes the human readable index and stuff , MAN page states that those arguments should be used when duplicating the VTOC from disk to disk
-) Create metaDBs on disk1 and mirror the disk
-) Create a filesystem on the new slice to use the additional space (of course on the metadevice)
-) D O N E

greets,
Christopher

Thanks for replies.

Nice hints joerg.

This was absolutely needed in procedure

but since disks are not identical I might do it manually, not sure yet.

Manually would be
# format
# partition
# print

get partition table from 73Gb drive

then with new 146Gb drive

# format
# partition
# (follow prompts etc.. to create partitions of 73Gb drive)

It says in the man fmthard "adds boot information to the Solaris fdisk partition"
so I do not think hint 2 is needed and if you partition manually the volume
manager will recreate the boot, swap, and other partitions.

Yes, the drives have reached there max capacity and backups of company
critical info have not been done in a month, it seemed like the best choice but
I wonder if there would have been a better?

I think it would be best to partition the drive exactly as mirror of 73Gb for volume
manager to do its thing. When both drives are in sync with 146Gb drives then
partion new space with format -> partition and mount a new file system. If you have
a suggestion or idea, I would like to here it.

I am pretty sure this should be used between same-geometry disks only.

And a hint: to step "0" - make a backup

Ooh, and what about the "metadevadm -u" after you swap drives?
This command updates mapping between so called "device ID" and cXtXdX identifier, and should be run (at least on Sol9) after disk replacements.

I agree, going to do it manually, safer.

Yes, but since the problem is no backups can be made, we have decided to
shutdown the oracle database when there is least trafic at 1:00 am (mind
you this may take about 2 hours) then pull the old drive and keep that
as the only backup so if we plug it in again it will start the oracle with a
clean startup.

Since I am using metadb -d (delete)
then metadb -a (attach) with new drive it should update database replica with new device id (mddb.cf file is updated with -a) but like it says in the metadevadm -u man "If the device ID has not changed this option does nothing", so it would do nothing probably but does not hurt to try it. The location of update would be in /etc/lvm/mddb.cf