On Mon, 25 Jun 2012 07:49:53 -0500
John Courcoul <
nospam....@removethis.me.com> wrote:
> So I was left with a SunFire V245 server that hosts several important
> services with two internal 146GB HDDs in RAID1 mirror config,
> containing the OS and applications and an external SAN with the data
> files. One of the two internal HDDs has failed and I need to replace
> it with a spare I have procured and is currently in an antistatic bag.
>
> As per the manual, the way to go about the replacement will be:
>
> 1. Put the new HDD in the chassis and have the OS recognize it.
> 2. Partition the new HDD identical to the failed unit.
> 3. Add the slices on the new as submirrors.
> 4. Detach the submirrors of the failed HDD.
> 5. Unmount any non-mirrored slices on the failed HDD.
> 6. Remove the dead HDD and use as a doorstop, paperweight, etc.
That's not quite how I would do it. This procedure creates a new set
of submirrors, which is rather complex, error-prone, and quite
unnecessary.
> Am I missing anything?
>
> What's the command used to add the new HDD to the system without
> having to do an INIT 6 and reboot the system?
Here's how I replace a failed SVM-only disk:
First, remove the dead disk, and plug the new disk in the bay you've
removed the dead disk from. The V245 supports hot-swapping.
Then, use the following command to copy the vtoc from the active disk:
# prtvtoc -s /dev/rdsk/c5t1d0s2 | fmthard -s - /dev/rdsk/c5t2d0s2
(assuming that the running disk is c5t1d0s2 and the new, empty disk is
c5t2d0s2).
Finally, use metareplace to sync the slices:
# metareplace -e themirror c5t2d0s0
etc.
> Do I need to format the slices and create UFS filesystems on them
> before attaching as submirrors, or do I just do the attach with
> METAREPLACE and let it synch itself?
No, fmthard takes care of partitioning, and metareplace only needs a
slice.
Take care,
--
Stefaan