1 |
> Or you can create raid0 with partitions on both drives (sda1+sdb1) |
2 |
> and "mirror" it on the second set of partitions on the same drives. |
3 |
> But then you do not have any protection from total disk failure |
4 |
> you'd normally expect from raid1. Only some very little protection |
5 |
> against a sector failure. |
6 |
> |
7 |
> Plus two equally strange configurations where you first create |
8 |
> raid1-mirrors and then strip them to raid0. Doing raid10 (or raid01) |
9 |
> with only 2 disks is imho not a good idea... |
10 |
|
11 |
It is very well written everywhere that deals with raid that it is a |
12 |
very bad practice to have more than one raid device per drive AND per |
13 |
bus. |
14 |
|
15 |
Having two identical disks and a need for redundancy calls for raid1 |
16 |
(mirror), nothing else. Or anything else is just fancy crap, but hey, |
17 |
you can learn a lot with fancy crap. |
18 |
|
19 |
If you wish to get something a little more than raid1, it might be with raid6: |
20 |
partition each drives in 2 equal sized partitions, then create the |
21 |
raid6 on those 4 devices. You loose both reading and writting |
22 |
performance due to seeking from one partition to another, but you keep |
23 |
redundancy since with raid6, 2 device failures can happen. The |
24 |
advantage here, is scalability: you buy a third drive later, |
25 |
partition it correctly and 'grow' your raid6 on those 2 new devices. |
26 |
But even then, if you buy a 3rd drive, i'd recommend raid5 simply on |
27 |
the whole disks (well one partition per drive). |
28 |
|
29 |
Here, do whatever scenario you want, and when you buy a 3rd disk: |
30 |
fail and remove the secondary drive on current raid |
31 |
create a raid5 with the new drive and the removed spare and 'missing' |
32 |
copy old raid over to new raid |
33 |
fail old raid and add its device to the new raid |
34 |
update your /etc files to point to the new raid |
35 |
|
36 |
Enjoy! |