Unix Tip: Mirroring your root partition with Solaris Volume Manager
Send in your Unix questions today! |
See additional Unix tips and tricks
Solaris Volume Manager can make easy work of mirroring your root file system, but you have to use the right commands in the right sequence to make easy work of this task. In this week's column, we'll run through each of the commands required to mirror root and show how you can check on what is happening in each step.
First, let's set the stage for our example system. We have a new Solaris 10 installation on a system with two internal disks and a disk array that we'll set up later. Each of the internal disks has been partitioned identically.
Part Tag Flag Cylinders Size Blocks 0 root wm 1452 - 23167 29.92GB (21716/0/0) 62737524 1 var wm 23168 - 24567 1.93GB (1400/0/0) 4044600 2 backup wm 0 - 24619 33.92GB (24620/0/0) 71127180 3 swap wu 0 - 1451 2.00GB (1452/0/0) 4194828 4 unassigned wm 0 0 (0/0/0) 0 5 unassigned wm 0 0 (0/0/0) 0 6 unassigned wm 0 0 (0/0/0) 0 7 reserved wm 24568 - 24619 73.35MB (52/0/0) 150228
The first thing we need to do is create the state database replicas. These are the data structures that store information about your volume configuration. We'll create multiple copies to ensure that we can recover if one copy becomes corrupt. You'll notice that we left a small partition for the state database replicas. We will now use the metadb -a -f command to create the replicas even if no database copies already exist. That -f option stands for "force".
# metadb -a -f c0t0d0s7 c0t1d0s7
In this example, we've only created two replicas, but you might want to create four or more to give you even better protection against potential problems. Database replicas can be created on dedicated slices or on slices that will become part of a concatenation or stripe.
To examine what we've done so far, we can use the metadb command without arguments:
# metadb
flags first blk block count
a u 16 8192 /dev/dsk/c0t0d0s7
a u 16 8192 /dev/dsk/c0t1d0s7
Now, we're going to use the metainit command to create a RAID 0 volume. Your choice of labels is flexible, but the convention we'll be using in this example is to name mirror components (or "submirrors") dx1 and dx2 and the mirrored device dx0 where "x" is a digit of 1 or more. Our root mirror will be d10, so our submirrors will be d11 and d12.
# metainit -f d11 1 1 c0t0d0s0 d11: Concat/Stripe is setup
The -f option is another "force" command. It tells metainit to create the submirror even if the slice contains a mounted file system.
Next, we need to create a RAID 0 volume on the second drive. We do this with a very similar metainit command, but we don't need the "force" option this time:
# metainit d12 1 1 c0t1d0s0 d12: Concat/Stripe is setup
If we check with metastat at this point, we'll see that our two submirrors have now been created.
# metastat
d12: Concat/Stripe
Size: 62737524 blocks (29 GB)
Stripe 0:
Device Start Block Dbase Reloc
c0t1d0s0 0 No Yes
d11: Concat/Stripe
Size: 62737524 blocks (29 GB)
Stripe 0:
Device Start Block Dbase Reloc
c0t0d0s0 0 No Yes
Device Relocation Information:
Device Reloc Device ID
c0t1d0 Yes id1,ssd@n20000004cfb583e8
c0t0d0 Yes id1,ssd@n20000004cfb57cd7
The next thing we need to do is configure the first of these two mirror components as a "one-way" mirror. This d11 submirror will then be the primary submirror.
# metainit d10 -m d11 d10: Mirror is setup
Looking at metastat output again, we see that d11 is now listed as a submirror of d10:
# metastat
d10: Mirror
Submirror 0: d11
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 62737524 blocks (29 GB)
d11: Submirror of d10
State: Okay
Size: 62737524 blocks (29 GB)
Stripe 0:
Device Start Block Dbase State Reloc Hot Spare
c0t0d0s0 0 No Okay Yes
d12: Concat/Stripe
Size: 62737524 blocks (29 GB)
Stripe 0:
Device Start Block Dbase Reloc
c0t1d0s0 0 No Yes
Device Relocation Information:
Device Reloc Device ID
c0t1d0 Yes id1,ssd@n20000004cfb583e8
c0t0d0 Yes id1,ssd@n20000004cfb57cd7
The metaroot command makes needed changes to your /etc/vfstab and /etc/system file. These changes will ensure that your system will mount the metadevices, not the underlying file systems, when it boots.
# metaroot d10
On Solaris 10, the changes required to the /etc/system files are minimal. See the line embedded between the two comments in the output below:
# tail /etc/system * set nautopush=32 * set maxusers=40 * * To set a variable named 'debug' in the module named 'test_module' * * set test_module:debug = 0x13 * Begin MDD root info (do not edit) rootdev:/pseudo/md@0:0,10,blk * End MDD root info (do not edit)
Changes to the vfstab file involve changing a line that looks like
this:
/dev/dsk/c0t0d0s0 /dev/rdsk/c0t0d0s0 / ufs 1 no -
to look like this instead:
/dev/md/dsk/d10 /dev/md/rdsk/d10 / ufs 1 no -Once you have gotten this far, it's time to reboot the system.# init 6 When the system is back up, it is time to attach the second submirror to the mirror. We do this with a metattach command. Notice that this isn't spelled "metaattach". # metattach d10 d12 d10: submirror d12 is attached With the second submirrored attached, the system builds the mirror of the root partition on the second submirror. Since, in this example, our root partition contains /usr, this part of the process will seem to be moving very slowly. We can,
however, check on our progress from time to time using this command: # metastat | grep -i sync State: Resyncing Resync in progress: 0 % done State: Resyncing Notice that, in this first check, we're 0% done. After a while, we will start
to see some progress: # metastat | grep -i sync State: Resyncing Resync in progress: 11 % done State: Resyncing It's going to take a while before significant progress has been made. Consider that a duplicate of the original slice is being created and that the status of our mirror is being updated as you monitor the slow progress. # metastat | grep -i sync State: Resyncing Resync in progress: 57 % done State: Resyncing At some point, we will try the metastat and grep command and notice that we're no longer getting output. At this point, the setup of the mirror is complete. # metastat | grep -i sync Next week, we will look at adding an alternate boot path so that we can boot off our mirrored drive as needed.