Edit /etc/xen/xend-config.sxp :
uncomment this line if you want to bridge your DomUs on the host :
# (network-script network-bridge)
uncomment/add this line if you want to use vnc without password by
default :
(vncpasswd '')
ignore the line wich say the bridge will be named xenbr0, it's not
true anymore, it now rename the interface eth0 in peth0 and create
a bridge named eth0
Reboot... tada!
:)
here's a sample from grub.cfg ( grub2 ) :
menuentry 'Debian GNU/Linux, with Linux 2.6.38-rc8-konradxen+ and XEN 4.0.2-rc3-pre' --class debian --class gnu-linux --class gnu --class os --class xen {
insmod part_msdos
insmod ext2
set root='(hd0,msdos5)'
search --no-floppy --fs-uuid --set 6149b0be-4294-49c5-871f-7195a3333cc8
echo 'Chargement de Linux 2.6.38-rc8-konradxen+ ...'
multiboot /xen-4.0.2-rc3-pre.gz placeholder dom0_max_vcpus=1 dom0_vcpus_pin
module /vmlinuz-2.6.38-rc8-konradxen+ placeholder root=/dev/mapper/vg0-root ro quiet
echo 'Chargement du disque mémoire initial ...'
module /initrd.img-2.6.38-rc8-konradxen+
}
verify it works by typing :
xentop
if it don't :
ls /proc/xen/
you should see :
capabilities privcmd xenbus
if not : add this line in /etc/fstab :
xenfs /proc/xen xenfs defaults 0 1
and execute
mount -a
if /dev/xen/evtchn doesn't exists add this line in /etc/modules
xen-evtchn
and execute
modprobe xen-evtchn
if there's no process xend
udate-rc.d xend defaults
/etc/init.d/xend/start
For DomUs : well it depends on what kind of DomU you want.
I didn't use a manager to create DomUs, but directly made the
config files ( you have to put them in /etc/xen ).
Here's two example :
pvops : /etc/xen/p2p.cfg
name = "p2p"
kernel = "vmlinuz-2.6.38-rc6-lio"
ramdisk = "initrd.img-2.6.38-rc6-lio"
root = "/dev/mapper/pvops-root"
memory = "512"
#disk = [ 'phy:vg0/p2p,sda,w' , 'phy:/dev/loop0,ioemu:hdc:cdrom,r' ]
disk = [ 'phy:vg0/p2p,xvda,w' , 'phy:vg0/mmd,xvdb1,w' ]
#boot = 'c'
vif = [ 'bridge=eth0' ]
vfb = [ 'type=vnc,vnclisten=0.0.0.0' ]
keymap = 'fr'
serial = 'pty'
vcpus = 2
full virtualized host ( the same DomU ) :
name = "p2p"
kernel = '/usr/lib/xen/boot/hvmloader'
builder = 'hvm'
device_model = '/usr/lib/xen-4.0/bin/qemu-dm'
memory = "512"
#disk = [ 'phy:vg0/p2p,xvda,w' , 'phy:/dev/loop0,ioemu:hdc:cdrom,r' ]
disk = [ 'phy:vg0/p2p,xvda,w' , 'phy:vg0/mmd,xvdb1,w' ]
boot = 'c'
vif = [ 'bridge=eth0' ]
vfb = [ 'type=vnc,vnclisten=0.0.0.0' ]
keymap = 'fr'
serial = 'pty'
vcpus = 2
If you want to install from an iso :
mount the iso on a loop device :
losetup /dev/loop0 /root/debian-sq-di-rc1-amd64-netinst.iso
add it to your config file (the commented line )
change boot='c' to by boot='d'
after the install remove it from the config and unmount it :
losetup -d /dev/loop0
to launch a vm :
xm create p2p.cfg
to hard stop it ( better to halt in the guest ) :
xm destroy p2p.cfg
Note for pvops :
You have to use a xen kernel and initrd that you put in /boot of
Dom0 ( you can use the one from Dom0 too )
You just have to copy /lib/modules/<kernel-version> into the
guest
As you can see i used lvm for disk devices the full path is
/dev/vg0/p2p, you can remove /dev for devices ( i m not sure it's
mandatory )
if you want to connect to the console change this line in inittab
:
1:2345:respawn:/sbin/getty 38400 tty1
by
1:2345:respawn:/sbin/getty 38400 hvc0
you will be able to see wat's hapenning by using '-c' option on
boot or 'xm console p2p' after
if you want to make lvm or partitions in the guest :
create a dos table and the partitions :
fdisk -l /dev/vg0/p2p
Device Boot Start End Blocks Id System
/dev/vg0/p2p1 1 61 487424 82 Linux swap / Solaris
/dev/vg0/p2p2 * 61 1045 7899136 8e Linux LVM
if you want to use debootstrap to install on those partitions :
here we use losetup again :
rmmod loop
modprobe loop max_part=63
losetup /dev/loop0 /dev/vg0/p2p
you will now have :
/dev/loop0p1
/dev/loop0p2
pvcreate /dev/loop0p2
vgcreate pvops /dev/loop0p2
vgscan
vgchange -ay ( maybe unecessary the first time because there no lv yet )
create your logical volumes and mount your logical volumes for use
with debootstrap
exemple :
mount /dev/pvops/root /mnt/chroot
cd /mnt/chroot
mkdir usr var dev proc home tmp sys
mount /dev/pvops/var /mnt/chroot/var
mount /dev/pvops/usr /mnt/chroot/usr
mount /dev/pvops/home /mnt/chroot/home
mount /dev/pvops/tmp /mnt/chroot/tmp
mount -o bind /proc /mnt/chroot/proc
mount -o bind /dev /mount/chroot/dev
mount -o bind /dev/pts /mount/chroot/dev/pts
mount -o bind /dev/shm /mount/chroot/dev/pts
mount -o bind /sys /mount/chroot/sys
install with debootstrap ( see manual/howto )
cp -rp /lib/modules/<kernel-version> /mnt/chroot/lib/modules/<kernel-version>
chroot /mnt/chroot
configure :
/etc/network/interfaces
/etc/fstab
/etc/hosts
/etc/resolv.conf
/etc/hostname
/etc/inittab ( for hvc0 )
exit
umount all your partitions
vgchange -an pvops
losetup -d /dev/loop0
and boot the vm :
cd /etc/xen
xm create p2p.cfg -c
Note for full virtualized host :
You must have a complete working system on the guest filesystem (
kernel, grub, etc, ... )
Note for both :
the disk devices created appears as xvda ,xvdb, etc
you can add a filesystem as a partition if you want : ( see xvdb1
in the config )
if you use vnc it will increment the port from 9001 as you add
VMs, you connect to it by using Dom0 IP
you call also use the extra options vncdisplay=10 ( for port 9010
)
you can set a vnc password for a specific vm : vncpasswd=idontknow
full exemple :
vfb = [ 'type=vnc,vnclisten=0.0.0.0,vncdisplay=10,vncpasswd=idontknow' ]
if you want to mount the same fs in different DomUs ( for
clustered filesystems ) add '!' in the config :
'phy:vg0/mmd,xvdb1,w!'
The end.
I thought it was short...
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
