HannibalWiki

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.

=====Debian domU=====

====Create stock rootfs====
Install the debootstrap package on the dom0 if you don't already have that package. Now create a minimal system installation in a subdirectory to bootstrap:
  apt-get install debootstrap
  mkdir -p /opt/xen/stock/deb-etch
  debootstrap --arch amd64  etch /opt/xen/stock/deb-etch  ftp://ftp.debian.org/debian
(this will use the /etc/apt/sources.list file on the dom0 in order to be able to find the Debian repository)

  chroot /opt/xen/stock/deb-etch
  apt-get install xfsprogs xfsdump udev vim ethtool

If (and ONLY if!) your dom0 has a x86-32 kernel/hypervisor, install TLS friendly libraries instead of disabling them in the domU (remember, in the old Sarge days we had to rename the dir /lib/tls to /lib/tls.disabled):
  apt-get install libc6-xen

Finally leave the chroot.
  exit

**TODO**

  Add more steps here to create the stock rootfs file, so it can be used during "Prepair a domU"








====Prepair a domU====

===kernel===
For administration maintenance copy the xen-kernel and its ramdisk from /boot in the dom0 to /opt/xen/kernels/<arch>

   mkdir -p /opt/xen/kernels/amd64
   cp /boot/vmlinuz-2.6.18-5-xen-amd64 /opt/xen/kernels/amd64
   cp /boot/initrd.img-2.6.18-5-xen-amd64 /opt/xen/kernels/amd64

As you want to use this kernel for your domU's, they'll need the kernel-modules, so copy them
from /lib/modules in the dom0 to /lib/modules/ in your domU rootfs

   cp -a /lib/modules/2.6.18-5-xen-amd64 /opt/xen/stock/deb-etch/lib/modules

===storage===
  * Create a volume for / (lvcreate -n <domU name>root -L <size> <volume group>

   lvcreate -n mydomuroot -L 1G vg00

  * Create a volume for swap

   lvcreate -n mydomuswap -L 200M vg00
   mkswap /dev/vg00/mydomuswap

  * tar or dd one of the stock rootfs files to the newly created volume. Remember to resize and check the filesystem afterwards if you chose to use dd.

   mkfs.xfs /dev/vg00/mydomuroot
   mount /dev/vg00/mydomuroot /mnt
   cd /opt/xen/stock/deb-etch
   tar cf - . | (cd /mnt && tar xf -)


===userland config===
adjust these files in the domU's root filesystem.
  * /etc/hostname (only hostname without domain)
  * /etc/fstab

      # /etc/fstab: static file system information.
      #
      # <file system> <mount point>   <type>  <options>       <dump>  <pass>
      proc            /proc           proc    defaults        0       0
      none            /dev/pts      devpts    defaults        0   0
      /dev/hda1       /               xfs     defaults        0       1
      /dev/hda2       none            swap    sw              0       0

  * timezone and localization (chroot /mnt tzconfig)
  * if required configure the network in /etc/network/interfaces and /etc/resolv.conf
  * configure mail delivery from local root (eg system message) to an existing address.

     echo "<your emailaddress> > /root/.forward
     install postfix and choose "satelite system"

  * install ssh server for remote access

     apt-get install openssh-server

  * as of debian-5.0 (lenny):
     apt-get install udev


===DomU config===
First create a subdirectory for the configuration files of your new guest.

   mkdir -p /opt/xen/mydomu

The configuration file itself /opt/xen/mydomu/domu.conf (also known as the spec file) has this content.
The guest in our example has 128Mb memory, one nic with a static MAC-address (we prefer to administer them ourselves), one root partition and one swap partition.

  kernel = "/opt/xen/kernels/amd64/vmlinuz-2.6.18-5-xen-amd64"
  ramdisk = "/opt/xen/kernels/amd64/initrd.img-2.6.18-5-xen-amd64"
  memory = 128
  name = "mydomu"
  vif = [ 'mac=00:16:3E:17:75:17,bridge=xenbr0' ]
  disk = [ 'phy:mapper/vg00-mydomuroot,hda1,w' , \
           'phy:mapper/vg00-mydomuswap,hda2,w' ]
  root = "/dev/hda1 ro"
  extra = "2"
  #extra = "console=hvc0 xencons=tty" # 2.6.26-x-xen

Now create a file to boot the new guest, /opt/xen/mydomu/domu.sh

  XEN_NAME=mydomu
  XEN_HOME=/opt/xen/${XEN_NAME}
  export XEN_NAME XEN_HOME
  echo "Starting XEN ${XEN_NAME} ..."
  /usr/sbin/xm create  ${XEN_NAME}  -f ${XEN_HOME}/domu.conf

Grant the proper permissions.

  chmod 755 /opt/xen/mydomu/domu.sh

Now you finally should be able to succesfully start the new guest.

  cd /opt/xen/mydomu
  ./domu.sh


===Automatic domU startup===

Create a directory /etc/xen/auto and put the domU configuration file in this directory (or create a symbolic link).

In case you have a lot of domU's, a problem might arise that networking fails when using automatic startup.

===extra software and config===
Configure locales to get rid of the message  Setting locale failed
  apt-get install locales
  dpkg-reconfigure locales

In case you get this message: "Couldn't get a file descriptor referring to the console", you might remove the console packages within the domU's as the dom0 already provides the keymap.
  apt-get remove --purge console-common console-data console-tools libconsole

To get rid of the en_US.UTF-8 warnings from apt/perl, add to /root/.profile
  unset LANG


To disable the indentation (which is by now the default in debian), add the two lines below to /etc/vim/vimrc:
  set noautoindent
  set paste


=====Opensolaris PV domU=====
In case your hardware does not match de Solaris-x86 hardware compatibility list, but you do want to run one or more Solaris machines, Xen is a solution. In this case, Linux is used for the dom0, so all drivers of Linux can be used to adress the hardware. The Xen hypervisor layer does the i/o translation :-)

Nexenta is a recently developed distribution of OpenSolaris, which combines the kernel and some great features like ZFS with the usability developed by Debian. A good direction for Solaris in my opinion.

The link below is describes the installation steps of the Solaris/Linux combination:
See: http://wiki.tardis.ed.ac.uk/index.php/OpenSolaris_Xen_domU


=====Ubuntu-10.04 PV domU =====

A convenent way to make the raw image is with KVM. Install from the Net or the CDrom, the usual way. After the reboot test, before the shutdown and copy-over to the Xen dom0, a few modifications are needed:

  * Install 10.04 with commandline only, and use expertmode.
  * Choose grub1 and not grub2. This is to have a normal /boot/grub/menu.list so pygrub will work without hacking it.

  * To have the xm-console, including a login prompt, do:
<code>
# most important:
cp  /etc/init/tty0.conf  /etc/init/hvc0.conf
sed -i 's|tty0|hvc0|g' /etc/init/hvc0.conf
# to be complete:
mv /etc/init/tty1.conf /etc/init/tty1.conf.disabled
mv /etc/init/tty2.conf /etc/init/tty2.conf.disabled
mv /etc/init/tty3.conf /etc/init/tty3.conf.disabled
mv /etc/init/tty4.conf /etc/init/tty4.conf.disabled
mv /etc/init/tty5.conf /etc/init/tty5.conf.disabled
mv /etc/init/tty6.conf /etc/init/tty6.conf.disabled
</code>

  * Optionally disable ureadahead:
<code>
mv /etc/init/ureadahead.conf  /etc/init/ureadahead.conf.disable
mv ureadahead-other.conf ureadahead-other.conf.disable
</code>


Now shutdown, and copy over the image file to the dom0 system. KVM usually leaves its image files in /var/lib/libvirt/images/*.img

Use dd to place this disk-image file on a logicalvolume.


  * Use this xm create config:
<code>
name = "ubu1004serv"
memory = 1024
vcpus = 1
#kernel = "/var/xen/kernels/ubuntu/vmlinuz-2.6.32-24-server"
#ramdisk = "/var/xen/kernels/ubuntu/initrd.img-2.6.32-24-server"
#root = "/dev/mapper/vg01-root ro"
#
bootloader = "/usr/lib/xen-3.4/bin/pygrub"
#extra = "console=hvc0 xencons=tty"
#
vfb = [  ]
on_poweroff = "destroy"
on_reboot = "restart"
on_crash = "restart"
#
disk = [ "phy:/dev/mapper/vg00-ubu1004disk0,xvda,w"  ]
vif = [ "mac=00:16:3c:53:ba:f9,bridge=xenbr0"]
</code>
HannibalWiki

User Tools

Site Tools

Page Tools
