Creation of metadevice

Here are the steps i used for creation of Metadevices for n-way mirror.

Assuming c0t0d0 is already set up, and c0t1d0 is not. You might want to save a copy of the vtoc if you want to.

prtvtoc /dev/rdsk/c0t0d0s2 | fmthard –s - /dev/rdsk/c0t1d0s2

we need (0.5n+1) replicas for the recovery to work so we need minimum 3 state database replicas per disk. creating in the last slice.

metadb -f -a -c 3 c0t0d0s7
metadb -a -c 3 c0t1d0s7

Creating the mirror for the root disk

metainit -f d10 1 1 c0t0d0s0
metainit d20 1 1 c0t1d0s0
metainit d0 -m d10

Configure the system to boot the root filesystem from the metadevice, using the "metaroot" command. This will make the necessary changes to /etc/vfstab and /etc/system:

metaroot d0

You may want to get the name of what is now the raw root disk, in case we need it later:

ls -l /dev/rdsk/c0t1d0s0

creating a 1-way mirror for the other slices. use the "-f" option to force the creation for the first mirror. You can create subsequent mirror slices using this method..

metainit -f d13 1 1 c0t0d0s3
metainit d23 1 1 c0t1d0s3
metainit d3 -m d13

Please modify /etc/vfstab so that the system knows how to load the disks..

#device  device  mount  FS fsck mount  mount
#to mount to fsck  point  type pass at boot  options
#
/dev/md/dsk/d3 /dev/md/rdsk/d3 /opt/ ufs 2 yes logging
/dev/md/dsk/d4 /dev/md/rdsk/d4 /var ufs 2 yes logging
/dev/md/dsk/d5 /dev/md/rdsk/d5 /usr/ ufs 2 yes logging

To be on the safe side, get the system to unlock file system and flush any UFS logging of the master file system before you reboot.

lockfs -fa
reboot

Now attach the secondary mirror to create 2-way mirror for s3, s4, s5

metattach d3 d23
metattach d4 d24
metattach d5 d25

After completion of the syncing process (you can monitor using metastat), you may want to reboot the server to check that all is well.

More reference:
http://bradthemad.org/tech/notes/disksuite_mirroring.php
http://www.sun.com/bigadmin/content/submitted/metadevices.html


---------------------------------------
Addition.

From Solaris 10 6/06 (Update 2) onwards, if you have a production server that root drive has not been mirrored and the root and swap take up the entire disk...

You may try to steal a few cylinders from swap to create the metadb but be in for a big surprise.. :) Solaris 10 does not allow you to do that... see below..

# swap -d /dev/dsk/c0t0d0s1
# format
Searching for disks…done
 
AVAILABLE DISK SELECTIONS:
0. c0t0d0 
1. c0t1d0 
Specify disk (enter its number): 0
selecting c0t0d0
[disk formatted]
Warning: Current Disk has mounted partitions.
/dev/dsk/c0t0d0s0 is currently mounted on /. Please see umount(1M).
/dev/dsk/c0t0d0s1 is currently used by swap. Please see swap(1M).
 
format> par
partition> 1
Part Tag Flag Cylinders Size Blocks
1 swap wu 0 - 429 989.34MB (430/0/0) 2026160
 
Enter partition id tag[swap]:
Enter partition permission flags[wu]:
Enter new starting cyl[0]:
Enter partition size[2026160b, 430c, 429e, 989.34mb, 0.97gb]: 414c
partition> 7
Part Tag Flag Cylinders Size Blocks
7 unassigned wm 0 0 (0/0/0) 0
 
Enter partition id tag[unassigned]:
Enter partition permission flags[wm]:
Enter new starting cyl[0]: 414
Enter partition size[141360b, 30c, 443e, 69.02mb, 0.07gb]: 16c
 
partition> l
Cannot label disk when partitions are in use as described.

haha.. here's the workaround i found on the web...

1) Start format
2) select your new mirror disk, in this case c0t1d0
3) Partition it like your current root disk, except you really want to set it up with a smaller swap and a new slice 7 for metadb. Leave all other slices alone.

Here’s the interesting bit…. you write the VTOC from our mirror disk to the root disk, usually you’d do it the other way….

# prtvtoc -h /dev/rdsk/c0t1d0s2 | fmthard -s - /dev/rdsk/c0t0d0s2

That’s it, you now have two identical VTOCs and a slice 7 for metadb’s... I think SUN is trying to reduce the number of incident where we mess up our systems??