raid10??

also sprach Héctor González [2008.01.30.0600 +1100]:
> Well, if memory doesn't fail, I just did something like:
>
> mdadm --create /dev/md0 --level=raid10 --raid-devices=4 /dev/sd[bcde]1

This has the advantage that access to the disk only has to traverse
the md layer once. It works, I use it almost everywhere.

--
.''`. martin f. krafft
: :' : proud Debian developer, author, administrator, and user
`. `'` http://people.debian.org/~madduck - http://debiansystem.info
`- Debian - when you have better things to do than fixing systems

die wahrheit ist selten
auf seiten der wahrscheinlichkeit.
-- heinrich v. kleist

Héctor González wrote:
> randall wrote:
>> Héctor González wrote:
>>
>>> With mdadm you can make RAID10, is there a problem here I just don't see?
>>>
>>> my mdadm.conf has this for one server with 4 500GB disks, md0 is 1 TB.
>>>
>>> ARRAY /dev/md0 level=raid10 num-devices=4
>>> UUID=9fbd86dc:518e4bf3:c0ae5f04:5c65c500
>>>
>>>
>>> Roberto C. Sánchez wrote:
>>>
>>>> On Tue, Jan 29, 2008 at 05:23:25AM +0100, randall wrote:
>>>>
>>>>
>>>>> i have just installed with the debian installer making 2 x raid1 and
>>>>> then glueing them together with LVM
>>>>> 256 MB /boot raid 1 (on all 4 disks)
>>>>> 10 GB /root raid1 + lvm
>>>>> 1 GB /swap raid1 + lvm
>>>>> 500 GB unused raid1 + lvm
>>>>>
>>>>> (i wasn't very sure what to do about the swap but i think this means
>>>>> slower but securer in a case of crashing)
>>>>>
>>>>> till so far the performance feels pretty snappy but i still have to do
>>>>> some benchmarking and remove a few random disks to see what happens.
>>>>>
>>>>> anybody ever reliably used a set up like this in production?
>>>>>
>>>>>
>>>>>
>>>> I run a similar setup on many production servers. On one, for example,
>>>> I have four disks. I create two RAID1 arrays, /dev/md0 and /dev/md1.
>>>> Then I create a LVM volume group, adding both /dev/md0 and /dev/md1 as
>>>> physical volumes. Then I create whatever logical volumes I want. It
>>>> works very well.
>>>>
>>>> Regards,
>>>>
>>>> -Roberto
>>>>
>>>>
>>>>
>>>
>> no problem, just some time left to try out different options ;)
>>
>> and i was just about to try that way, any pointers about the install you
>> are willing to share?
>>
>> should be something as follows i figured, can you include /boot this way?
>>
>> sfdisk -d /dev/hda | sfdisk /dev/hdb etc......
>>
>> mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sd[ab]1
>> mdadm --create /dev/md1 --level=1 --raid-devices=2 /dev/sd[cd]1
>> mdadm --create /dev/md2 --level=0 --raid-devices=2 /dev/md[01]
>>
>>
>>
>>
> Well, if memory doesn't fail, I just did something like:
>
> mdadm --create /dev/md0 --level=raid10 --raid-devices=4 /dev/sd[bcde]1
>
> I don't boot currently from that raid, but it should work fine, if you
> told the mdadm package to include boot raid support.
>

almost there ;)

i tried to follow the how-to-forge article on raid1 installation for a
running system, replacing it with raid10.
this seems to work nicely, attached my notes based on the original article.
until the very last final step that is, getting the final raid10 to
boot, i'm not having much luck with grub, i'm receiving grub error 2 at
the moment for all 4 disks.

before i continue on this path i want to ask if it is possible at all.
md0 /root
md1 /swap
md2 /data

so all 3 as RAID10 and no seperate /boot

so here's the question.
1 this is possible and am i just to big a noob to get the correct grub
parameters
2 this is not possible so i should use a seperate /boot in raid1