On 4/18/2011 11:08 AM, Hans de Goede wrote:
Hi,
On 04/18/2011 04:37 PM, Bill Nottingham wrote:
Hans de Goede (hdegoede@xxxxxxxxxx) said:
I'm not sure this is a good idea. I'm getting a bit rusty wrt
RAID things now, but IIRC, we need to write out an mdadm.conf with
"AUTO +imsm +1.x -all"
In there, even on a system without any arrays, to ensure that any
Intel BIOS RAID arrays which are created on a system (from the BIOS)
post install, will get properly activated.
All in all my 2 cents are that this fix is a bit risky this late
in the cycle. Esp. when there is a much easier low risk fix 696907
available in the form of making initscripts check that /sbin/mdadm
exists before trying to execute it.
TBH I don't see how this is an anaconda bug at all.
I think it's somewhat questionable for anaconda to write a mdadm
configuration file while at the same time not installing mdadm.
How to handle that can certainly be debated, though.
Good point, perhaps we should not allow installing without mdadm
at all, since it now comes with udev rules for things like
firmware raid activation. Do we allow installing without dmraid ?
Either one seems like a bad idea (as we could end up using single
discs of a mirror without marking it out of sync which is BAD).
I had always assumed that mdadm was in the same class as udev as
something that is always present. If the kernel has md.ko then
userspace should have mdadm.
--
Dan
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
