Matt said:
If I have an older system running a RAID1 that stores my pictures, music
etc... and the motherboard dies will a new system recognize the data on the
drives or would I have to somehow back up the existing RAID and reinstall
everything from scratch?
Since you have to back up the data anyway, why worry ?
You may be thinking, that RAID 1 makes a bulletproof hardware
setup, like you were on the Space Shuttle or something. Consider
the power supply 12V rail, rises to 15V while you're out in the
kitchen making coffee. When you come back, both your RAID 1
drives are burned and inoperable. (Someone on the groups,
confirmed that's exactly what happened to him, power supply
failure.)
That's why, even with RAID 1, you've got a backup disk, with
"bare metal" recovery capability as a feature of your
backup software. That way, you can restore the information
on another computer and another disk.
A Repair Install can be used, once you've done the basic restore
to the new computer, to prepare your OS for operation on the new
hardware. If a new driver is required, you can press F6 and install
it and so on.
The layout of RAID 1, isn't necessarily that fancy. It might look
like this, on each disk. The metadata might be up near the end, out
of harm's way. When the RAID software prepared the disk, it was
careful to trim the "claimed" disk capacity, so that no overwrites
of the metadata could occur. The metadata identifies that the disk
is part of an array, whether it's the odd or even disk, the
stripe size (or similar storage size parameter) and so on.
+------------+-------------+---------------+
|MBR | Partitions | RAID metadata |
|(the usual) | | }
+------------+-------------+---------------+
For a disk like that, it might well work, when plugged alone
into a new computer (ignoring the potential for driver errors).
But I've seen other cases, where the metadata is somewhere else.
I had a Promise based setup once, where the first partition used
to "disappear" when the drive was connected to another controller.
The other three partitions were still visible. So there are other
options available.
If you're really curious, you can do like I did, and actually zero
out the disks, fire up the RAID software, create an array, then
check the entire disk to find where the metadata is being stored.
I found a 64KB block, about a megabyte or two from the end of the
disk, on each hard drive in the array. It contained a structure
capable of recording up to 16 array configurations.
A striped array (RAID 0) is harder to move. To recover a stripe,
you'd need to de-interleave the two disks, to build a single disk
version of the same thing. RAID 1, by comparison, is a bit more
straight-forward. But learning more about how your RAID works,
never hurts.
It's one of the things I recommend to RAID users. Do all sorts
of experiments, when the disks have no useful data on them. On
a RAID 1, unplug one disk and see what happens. Take the removed
disk, and zero it on another computer (to make a "blank" replacement
disk). Bring it back to the RAID 1 computer, and do a "rebuild".
Did it work ? And when you check the contents of the two volumes,
are they identical ? At least one SIL3112 owner, discovered when
it was too late, that the hardware had stopped "mirroring" about
three months before one of his disks failed. It pays to learn
as much as possible about your disks, before valuable data
(with no backups) is on there, and you get a warning or
error some day, while booting.
Paul
