Well yes, I had previously tried the lines that were commented out,
including the DEVICE statement as listed.  And I had followed one of the
conventions given in mdadm.conf(5) -- using the "*" to wildcard the
devices.  I had seen other examples on the net where an SA had used
multiple DEVICE statement to separate out the devices with their
respective ARRAYs, but there's no mention of that being needed in the
man pages of mdadm/mdadm.conf that I could find.

And I may be a little dense here, but I still don't see an "mdadm
--assemble" in the startup anywhere, so I'm a little confused as to how
this is supposed to happen.  Rough being a raidtools kinda guy for all
those years...  ;)

Thanks!
Rich

-----Original Message-----
From: Patrick Lauer [mailto:[EMAIL PROTECTED] 
Sent: Wednesday, November 23, 2005 9:38 AM
To: [email protected]
Subject: Re: [gentoo-server] Auto allocating RAID sets on reboot w/mdadm


On Wed, 2005-11-23 at 09:21 -0600, Jesse, Rich wrote:
> Hey all,
> 
> I've got two RAID-0 sets (I know, I know, but they really are
expendable
> and it's my test box) but when I reboot, the RAID sets are not
> reallocated.  I've looked in /etc/init.d/mdadm and
> /lib/rcscripts/raid_start.sh, but "mdadm --assemble" isn't called and
> consequently the RAIDs  aren't mounted.  Here's the pertinent lines
from
> /etc/mdadm.conf:
> 
> #DEVICE       /dev/sd*
> #ARRAY        /dev/md0 devices=/dev/sda,/dev/sdb,/dev/sdc,/dev/sdd
> #ARRAY        /dev/md1 devices=/dev/sde,/dev/sdf
> ARRAY /dev/md0 level=raid0 num-devices=4
> UUID=bb968c49:5ca8ea1b:383ac76f:8950b88a
> ARRAY /dev/md1 level=raid0 num-devices=2
> UUID=96ba9471:e161224f:6bac7812:23c3f914
add a DEVICE line ...
mdadm will only use disks that are listed there, e.g.
DEVICE /dev/sd[abcd]1 

[snip]
> And Google hasn't really helped.  The Gentoo Wiki's great for
> installation and setup, but I wasn't able to find anything about this.
man mdadm / mdadm --help
> Thoughts?  This must be really simple...
It usually is :-)

wkr,
Patrick
-- 
Stand still, and let the rest of the universe move

-- 
[email protected] mailing list

Reply via email to