[sclug] Setting up SATA RAID for /home not / (Debian Lenny)

Dickon Hood dickon-ml at fluff.org
Wed May 20 15:35:58 UTC 2009


On Wed, May 20, 2009 at 16:09:15 +0100, John Stumbles wrote:
: Keith Edmunds wrote:

: ># mdadm -C /dev/md0 /dev/sdc1 /dev/sdd1 -n 2 -l 1
: ...
: ># mke2fs -j /dev/md0 (or whatever filesystem you want)
: ># mount /dev/md0 /it/goes/here

: I know it's just cosmetic, but is there a Linuxey 'standard' sort of 
: place to mount a filesystem that's going to be exported (this box is a 
: fileserver)?

Not really.  The BBC uses /content; I use /space -- it doesn't really
matter.

: >Add /dev/md0 to /etc/fstab if desired.

: OK, thanks, and thanks Dickon:

: > and you'll probably want to set the partition type to 0xfe (RAID
: > autodetect).  If you're using a Debian kernel, you'll probably want to

: > mdadm --misc --scan --brief > /etc/mdadm/mdadm.conf

: When I installed mdadm it asked me:

: "Once the base system has booted, mdadm can start all MD arrays (RAIDs) 
: specified in /etc/mdadm/mdadm.conf which have not yet been started. This 
: is recommended unless multiple device (MD) support is compiled into the 
: kernel and all partitions are marked as belonging to MD arrays, with 
: type 0xfd (as those and only those will be started automatically by the 
: kernel)."

AIUI that's if it's built-in to the kernel, and not loaded as a module in
an initrd (which it almost certainly will be if you're using a distro
kernel).

: which sounds as if I don't need to set the type 0xfd ... we will see, I 
: haven't restarted the system yet, mke2fs is still cooking ...

If they're specified in mdadm.conf, you shouldn't need to type them.  I
prefer to.

It might be worth looking at ext4, BTW.  It's quite a lot more efficient
on large files.

-- 
Dickon Hood

Due to digital rights management, my .sig is temporarily unavailable.
Normal service will be resumed as soon as possible.  We apologise for the
inconvenience in the meantime.

No virus was found in this outgoing message as I didn't bother looking.



More information about the Sclug mailing list