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

Re: [Xen-users] domU on aoe troubles (saved by losetup?)


  • To: "Tom Brown" <tbrown@xxxxxxxxxxxxx>
  • From: "Chris Fanning" <christopher.fanning@xxxxxxxxx>
  • Date: Wed, 5 Sep 2007 17:52:03 +0200
  • Cc: Steven <xen@xxxxxxxxxxxxxxxxxxx>, xen-users@xxxxxxxxxxxxxxxxxxx
  • Delivery-date: Wed, 05 Sep 2007 08:52:48 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=OG5UTAEHFl1dowHOQq6WLc38zmDXbAa6Z1pQK7inVbqBXi6sCEVMO2ajV/dzZsqGqc/nL3+0nKIozNv/9mrinPEi8WQzOIbRoznw6pV6g00Rgim1PKnV98Y+OwsvGE7qnGRmPH8OohVxkUmVHCHh6nDMiAGexNG9VL2LS8pD0/Q=
  • List-id: Xen user discussion <xen-users.lists.xensource.com>

Hi,

I ended up using LVM to wrap AoE.

xen1# vi /etc/lvm/lvm.conf
scan = [ "/dev/etherd" ]

xen1# pvcreate /dev/etherd/e0.0
xen1# vgcreate vg_domU_root /dev/etherd/e0.0
xen1# lvcreate -L 5g -n lv vg_domU_root

but, on booting the machine, AoE devices were not available when LVM
started so I did an ugly hack. This could be done better modifying the
services rc links.
xen1# vi /etc/init.s/rc.local
case "$1" in
    start)
        do_start
        /etc/init.d/lvm start
        ;;
    restart|reload|force-reload)
        /etc/init.d/lvm restart
        echo "Error: argument '$1' not supported" >&2
        exit 3
        ;;
    stop)
        /etc/init.d/lvm stop
        ;;
    *)
        echo "Usage: $0 start|stop" >&2
        exit 3
        ;;
esac

Cheers.
Chris.

On 9/1/07, Tom Brown <tbrown@xxxxxxxxxxxxx> wrote:
> On Sat, 1 Sep 2007, Steven wrote:
>
> > Chris Fanning wrote:
> >>  Hello everybody,
> >>
> >>  I'm a bit stuck :(
> >>
> >>  I am trying to setup diskless xen servers.
> >>  I can boot dom0 via nfs and setup swap on nbd. that's ok.
> >>
> >>  Now I'm trying to use aoe for the domU's.
> >>
> >>  When I try to use the etherd devices the domU boot hangs when mounting
> >>  root filesystem
> >>  /etc/xen/domU.config
> >>  disk = ['phy:etherd/e0.0p1,hda1,w']
> >>
> >>  But if I do
> >>  losetup /dev/loop0 /dev/etherd/e0.0p1
> >>
> >>  and modify /etc/xen/domU.config accordingly
> >>  disk = ['phy:loop0,hda1,w']
> >>
> >>  then domU boots! :)
> >>
> >>  I was wondering why I can't use etherd/e0.0p1 directly.
> >>  What would be a good way to deal with this? with loop devices?
> >
> > Did you find out why and how to fix it?
> > I use aoe and domU hangs too when I select etherd/e0.0p1 directly.
>
> AFAIK you need to use an initrd file to boot directly off of AOE ...
> partly because you need to slow down the boot process a bit to let
> auto-discovery take place.
>
> You might also get away with specifying IP config parameters for your
> interface, as that _might_ cause it to be up instead of the default down
> (which will obviously kill aoe).
>
> My changes to the initrd scripts are basically
>
>     /bin/busybox ifconfig eth0 up
>     echo hello world > /dev/etherd/discover
>     sleep 5
>
> I also build some devices as I've had mixed experiences with udev
> doing what I want, but clearly "ifconfig eth0 up" is important...
>
> Anyhow, I/we use the above process on production domUs. The
> advantage to doing it this way is that you can easily migrate the
> domains... and if you are using vblade for an AOE server, and
> DRBD to mirror the underlying storage, you can actually move the
> vblade process around... the domU kernel very quickly picks up
> the new MAC address :) ... now I just need to get better at
> moving drbd processes around, as I need to simplify some things
> here.
>
> -Tom
>
>

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.