Installing Eucalyptus (1.6) on Debian "squeeze"

Eucalyptus can be installed on Debian squeeze using binary DEB packages. Squeeze has not been released yet, so things can change quickly and without warning.

Download DEBs

Eucalyptus binary installation is broken up into several packages: one for each of the components (CLC, Walrus, CC, etc.), as well as a couple of common packages. To install them, along with a significant number of dependencies, add our repository to the list of repositories for your system to use. To do so, add somewhere in /etc/apt/sources.list file the following line:

For 1.6.1:

deb http://eucalyptussoftware.com/downloads/repo/eucalyptus/1.6.1/debian/ squeeze contrib

For 1.6.2 (including release candidates):

deb http://eucalyptussoftware.com/downloads/repo/eucalyptus/1.6.2/debian/ squeeze main

And then run:

apt-get update

After installation you may remove the entry from sources.list if you don't want to update Eucalyptus packages automatically.

Prerequisites

If you start with a standard Debian squeeze installation, you will satisfy all Eucalyptus prerequisites with the following steps:

Synchronize clocks (e.g., using NTP: ntpdate pool.ntp.org) across all Eucalyptus machines and client machines.

If using a firewall, permit the Eucalyptus components to communicate with one another, and permit clients to communicate with Eucalyptus.
- NOTE: on the front-end, ports 8443, 8773, 8774 and 9001 must be open. On the node, port 8775 must be open.

Ensure that locales are properly configured (use dpkg-reconfigure locales).

Install libvirt/qemu-kvm and configure it to run as user 'eucalyptus':

sudo apt-get install libvirt-bin qemu-kvm
# set the field user to be: user = "eucalyptus"
sudo vi /etc/libvirt/qemu.conf
# restart libvirt
sudo /etc/init.d/libvirt-bin restart

If running in SYSTEM networking mode, which is the default, your node machine(s) must be configured with a bridge as the primary interface. For example, you may try:

sudo apt-get install bridge-utils
sudo vi /etc/network/interfaces

Comment out any entry for your existing interfaces (eth0, eth1, etc) and add a bridge entry with your interfaces attached. For example, to have your bridge come up with all physical Ethernet devices added to it, and have DHCP assign an address to the bridge, use:

auto br0
iface br0 inet dhcp
      bridge_ports all

For a static configuration with just eth0 attached (substitute your actual network parameters):

auto br0
iface br0 inet static
      address 192.168.12.20
      netmask 255.255.255.0
      network 192.168.12.0
      broadcast 192.168.12.255
      gateway 192.168.12.1
      dns-nameservers 192.168.12.1
      dns-search foobar foobar.com
      bridge_ports eth0

Finally, restart the network by either by restarting the network with '/etc/init.d/networking restart' or by rebooting the machine.

Install DEBs on the front end

On front end, where cloud controller, Walrus, cluster controller, and storage controller will run, install the appropriate DEBs:

aptitude install eucalyptus-common eucalyptus-cloud eucalyptus-walrus eucalyptus-sc eucalyptus-cc

Install DEBs on the nodes

On the compute nodes, install the node-controller DEB:

aptitude install eucalyptus-nc

(You may safely ignore the error adduser: The group 'libvirtd' does not exist.)

Firewall - how exactly?

Submitted by smoe on Sat, 05/29/2010 - 11:57.

Hello,

I needed to communicate the ports and the protocols to our admins and felt a bit lost. All nodes get on some dedicated subnet, and they would see a different set of rules to access to front node than a visitor from outside would. Also, the access to the nodes from the outside will differ from the access of the front node from the outside.

Please elaborate a bit about what ports need to be open for euca2ools (the outside, i.e.collaborators) and for node-host communication. Many thanks

Steffen

Please submit your questions

Submitted by graziano on Wed, 06/02/2010 - 09:41.

Please submit your questions to the forums: you will get faster answers and it keeps the documentation easier to follow.

the loop module

Submitted by smoe on Sat, 05/29/2010 - 11:40.

A message on missing loop devices may appear. This is indicatory of a missing loop module. If
"modprobe loop max_loop=256" does not work, then as root

# you may have this one already
apt-get install module-assistant

# get, build and install the loop module
m-a a-i loop-aes-source

I am not sure if this only installs the Debian package (don't think so) or if it also installs the kernel module by calling modprobe. To be on the safe side, do
"rmmod loop; modprobe loop max_loop=256"

Steffen

You are installing the AES

Submitted by graziano on Wed, 06/02/2010 - 09:40.

You are installing the AES encrypted loop device: as far as I know the standard Debian kernel comes with loop modules so the instructions should work. If your kernel has the loop device compiled in, you need to pass max_loop=255 on the kernel boot command line.