Adding spare drive

Adding spare drive

[mousebusiness.com]

I will be building a RAID5 array with 5 data drives and one hot spare. I
need to use the drive for hot spare for something else at the moment. Is it
possible to build my array with 5 drives and add hot spare later by simply
adding the entry for it in /etc/raidtab?

The array will be built with 6 drives connected to 3 IDE controllers, one
per channel, 2 channels on each board. If one controller fails, obviously
the array will go down with it. Does that mean that all data will be
irrecoverably lost or will things go back to normal after I replace
defective controller and let the array resync?

I know these are basic questions, but I didn't find answers in FAQ (hint,
hint)

Comments greatly appreciated.

Artur

Re: Adding spare drive

[Derek Vadala]

On Mon, 6 May 2002, mousebusiness.com wrote:

> I will be building a RAID5 array with 5 data drives and one hot spare. I
> need to use the drive for hot spare for something else at the moment. Is it
> possible to build my array with 5 drives and add hot spare later by simply
> adding the entry for it in /etc/raidtab?

Just create a 5-disk RAID-5 without a hot spare. When you're ready to
bring the spare disk into the array just use raidhotadd to insert it.

# raidhotadd /dev/md0 /dev/<sparedevice>

Now <sparedevice> is a spare disk in /dev/md0 (the 5-disk RAID-5). Now
update /etc/raidtab to reflect the new config. That means you'll add
nr-spare-disks and a device/spare-disk entry for /dev/<sparedevice>.

You can raidstop and raidstart the array to make sure it works, if it
makes you feel better, but you shouldn't need to.

> The array will be built with 6 drives connected to 3 IDE controllers, one
> per channel, 2 channels on each board. If one controller fails, obviously
> the array will go down with it. Does that mean that all data will be
> irrecoverably lost or will things go back to normal after I replace
> defective controller and let the array resync?

You'll be able to recover from a controller failure by replacing the
failed controller. Since physical controller placement typically impacts
device names you want to make sure you put a new controller in the same
slot as the one the failed, or make sure you have mapped out the disk
order and changed /etc/raidtab to reflect any changes.

---
Derek Vadala, derek@cynicism.com, http://www.cynicism.com/~derek