[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: software raid on /



On Saturday 09 September 2006 19:45, Michael Ott wrote:
> Hi!
>
> I have an box with raid1 for every partitions. But since a replace of a
> harddisk the / partition will only use one disk. I can add the other
> part using mdadm.
>
> Here my menu.lst
> title           Debian GNU/Linux, kernel 2.6.8-3-686
> root            (hd0,0)
> kernel          /vmlinuz-2.6.8-3-686 root=/dev/md1 ro
> initrd          /initrd.img-2.6.8-3-686
> savedefault
> boot
>
> Using standard kernel. What I have to change?


I am guessing, because I don't know exactly what it is, but

initramfs requires that the use of raid is switched on.  In 
particular /etc/mdadm/mdadm.conf should have some records in it that 
list define the arrays, and then update-initramfs gets them set up.  Mine is 
like this

DEVICE partitions
CREATE owner=root group=disk mode=0660 auto=yes metadata=1
MAILADDR root
ARRAY /dev/md0 level=raid1 num-devices=2 name=0 
UUID=c4887a6b:db58843e:209f5cfe:35d78da6
ARRAY /dev/md1 level=raid1 num-devices=2 name=1 
UUID=210a27fe:06a1f486:1907fc2f:b4172f32
ARRAY /dev/md2 level=raid1 num-devices=2 name=2 
UUID=1508c72f:91e305da:a71136c7:f7b25b0b
ARRAY /dev/md3 level=raid0 num-devices=2 name=3 
UUID=9df2ce56:80fd9cd9:6265d55b:9dddda0b
ARRAY /dev/md4 level=raid5 num-devices=3 spares=1 name=4 
UUID=9ade8b08:39950784:df2dcf3f:18e8cfa9


I don't know exactly how it works when you replace a failed drive, but I think 
(inside the initramfs) mdadm looks for disks with the UUID in the metadata 
someplace.  

So I think the process is probably to start your system and then add the new 
disk that replaced the broken one with mdadm -A 


-- 
Alan Chandler
http://www.chandlerfamily.org.uk



Reply to: