solaris svm and raid5 : a way for expand on fly?

Question

I know two methods for expand raid5 on solaris svm using UFS one is this another is to replace disk by disk,suppose i want to remove old small disk and replace with bigger disk using this procedure

devfsadm
cfgadm -c configure sata2/0
format -d c0t5d0
metadb -a -f c0t5d0s2
metareplace -e myraid c0t4d0s2 c0t5d0s2
metadb -d c0t4d0s2
cfgadm -c unconfigure c0t4d0s2

I have replaced all disks with the method above and my raid5 is online and ok as metastat said But after give

metadevadm -vr

and

growfs -M /raid /dev/md/rdsk/d44

The SIZE is the same as raid with old disks and this is wrong because i replaced disks with bigger disks. On linux is really easy to replace raid5 disk on fly and grow the raid5 (mdadm fail,add,grow,then pvresize..) on fly,i miss something on solaris svm? The first method is also good(concatenate+growfs) but i want to replace disks old(small) with new(big). Please don't answer zfs,for "study" reason i'm on ufs+svm

Answer 1

It's been awhile since I've had to touch SVM, but I'm inclined to agree with BitsOfNix that it's not possible to do what you are trying to do. Memory is that you can only do what you're trying to do with RAID 1 on SVM.

Did you verify that metastat sees that the blocksizes have changed for the larger disks and the "RAID device?"

I think you're also missing a step prior to your growfs. Memory is that you need to expand the slice of "RAID device" prior to the growfs. And with RAID1 on SVM, then a metasync for the mirroring? You might need to do something similar for RAID5 on SVM.

You mention doing this for a "study," but I'm not sure it's fair to ding SVM compared to mdadm since SVM was created and used in a different time to meet a different need.