On Sun, Nov 16, 2008 at 04:14:38PM +0200, Cem Kayali wrote:
Hi,
First of all, i did my homework, and tried various configurations and
advises in (NetBSD's) mailing-lists and other net resources.
I can not boot Xen domU linux. Here are configurations:
1- I partitioned hard disk using GParted. Partition 1: Ubuntu x86/64.
Partition 2: Shared swap. Partition 3: NetBSD 5.99.01 amd64. Partition
4: Empty now, there will be another OS.
2- I have installed Xen as described in NetBSD's how to, as dom0. I used
NetBSD's multiboot support and there is no problem with NetBSD as dom0.
It boots perfectly.
3- Best /usr/pkg/etc/xen/l?nux configuration so far:
kernel = "/vmlinuz-2.6.24-21-xen"
memory = 512
name = "linux"
cpu = "^1"
vif = []
disk = [ 'phy:/dev/wd0e,hda1,w' ]
root = "/dev/hda1"
Problems and how i fixed some of them:
1- I always see "vif hotplug error", regardless whatever i use as vif=[]
value. I postponed to f?x this error, and used "xenstore-write
/local/domain/0/backend/vif/8/1/hotplug-status connected" command.
Yes, I think you need to associate it to a bridge (or a local IP if using
a routed config instead of bridged). Otherwise the vif script is not run,
and it's this script which toggle hotplug-status.
2- Next problem was "vbd hotplug error". It took my several hours to
notice that this is because of AHCI controller. I modified bios settings
and switched to IDE controller and it plugged hard disk drive. Funny
thing is that both OS support ahci controller without problem --- but
not xen guest kernel.
I don't understand how AHCI would make a difference, exept maybe if the
drive are probed in a different order (this happens with Intel controllers
at last, when switching between compat and AHCI modes). Did you check that
you could access /dev/wd0e from dom0 (something like
dd if=/dev/rwd0e of=/dev/null) ?
3- Now current problem... I couldn't fix.... I see "Xenbus initializing
devices message". It counts down starting from 200+ seconds to 0 and
gives up... Last message in screen is "could not mount root (3,1)
partition"... I used 0x301 instead od hda1 to fix this issue and still
same. I also tried mounting volume to /mnt first and then use root="/mnt
ro" too. xend.log shows:
Do you see anything relevant to the disk device on the domU's console
when the linux kernel boots ? At last with fedora and RHEL/RHEL derivatives,
xen drivers are not present in the kenrel and are loaded as module, so
the kernel has to be booted with a ramdisk to be able to load the xbd
driver and switch to the real root.
It's also possible that the disk device doesn't show up as hda1 but xvda1