> I am running 2.2.12. What I don't understand is if both md0 and md1 are
> kicked out, why am I still able to use md2. I can read information
> without any problems at this point. That seems strange.
Are you sure it's not just information in the disk cache? the disk cache
can grow to be many megabytes...
> Will recreating the mirror device, md2, cause any data loss?
Nope. It's not like mke2fs, it doesn't overwrite any of your data, just the
raid superblocks. You do have to make sure that your /etc/raidtab is up to
date.
> Reason I'm not running raid 5 is because I was under the impression that
> with this configuration, a raid 1/0 config, I was providing more
> redundency with the advantages of a stripped array for performance. Can
> you give me reasons why raid 5 would be better other then retaining more
> available space. My goal isn't necessarily space, but more for
> redundency.
The problem is that this configuration only provides more redundancy in some
situations. i.e., you have to get "lucky" on the second disk failure. If
/dev/sdc and /dev/sdd both failed, your whole array would be down. With one
drive down, and an even statistical distribution of failures... this
configuration only gives you a 40% chance of surviving a 2nd disk failure.
You could run raid5 with a spare disk. As soon as the 1st disk fails, your
second will begin rebuilding... limiting the amount of time you are open to
a 2nd failure (though this doesn't guard against forms of bus or power
failures that will take more than one drive at once).
If you have to survive more than one failure at once, you'll need to lose a
little more disk space. Running raid1 on each of three disk pairs, and then
running raid5 on the result would allow you to have and 2 drives fail, and
possibly several more. You'll end up with 2 disks worth of space instead of
3, but you'll be guaranteed 2 failures won't kill you.
In my opinion, I'd only use one side or the other... either the single raid5
that will support 1 disk failed, or the super paranoid 1/5 raid that will
support two.... But I wouldn't use something that only gave me a 40% chance
of surviving the 2nd disk failure.
Tom
