This is a puppet module to configure /etc/network/interfaces for OpenStack deployments. The networking::interfaces class supports the following node types:

A. controller

B. compute

C. swift-proxy

D. swift-storage

E. load-balancer

Here is a sample compute node definition that calls the networking::interfaces class. This example uses VLANs (100) to seperate the private (Nova Flat) network from the public/management network.

# Puppet node definition
node /<compute_node>/ {

  class { 'networking::interfaces':
  # Node Types: controller, compute, swift-proxy, swift-storage, or load-balancer
    node_type           => compute,
    mgt_is_public       => true,
    vlan_networking     => true,
    vlan_interface      => "eth0",
    mgt_interface       => "eth0",
    mgt_ip              => "192.168.10.10",
    mgt_gateway         => "192.168.10.1",
    flat_vlan           => "100",
    dns_servers         => "192.168.254.254",
    dns_search          => "myco.com",
  }
}

Here is a sample compute node definition that calls the networking::interfaces class. This example uses physcial interface (eth0 and eth1) to seperate the private (Nova Flat) network from the public/management network.

# Configure /etc/network/interfaces file
class { 'networking::interfaces':
# Node Types: controller, compute, swift-proxy, swift-storage, or load-balancer
node_type           => compute,
mgt_is_public       => true,
mgt_interface       => "eth0",
mgt_ip              => "192.168.220.53",
mgt_gateway         => "192.168.220.1",
flat_interface      => "eth1",
dns_servers         => "192.168.220.254",
dns_search          => "dmz-pod2.lab",
 }
}

Here is a sample controller node definition that calls the networking::interfaces class. This example uses physcial interface (eth0 and eth1) to seperate the private (Nova Flat) network from the public/management network. The primary difference between compute and controller is assigning an IP address to the flat interface.

# Configure /etc/network/interfaces file
class { 'networking::interfaces':
# Node Types: controller, compute, swift-proxy, swift-storage, or load-balancer
node_type           => compute,
mgt_is_public       => true,
mgt_interface       => "eth0",
mgt_ip              => "192.168.220.53",
mgt_gateway         => "192.168.220.1",
flat_interface      => "eth1",
flat_ip             => "<ipaddrofnovaflat_net>"
#add flat_mask if the net is not a /24
dns_servers         => "192.168.220.254",
dns_search          => "dmz-pod2.lab",
 }
}

Here is a sample load-balancer node definition that calls the networking::interfaces class. This example uses physcial interface (eth0) as the management and interface used for load-balancing traffic.

# Configure /etc/network/interfaces file
class { 'networking::interfaces':
# Node Types: controller, compute, swift-proxy, swift-storage, or load-balancer
node_type           => load-balancer,
mgt_is_public       => true,
mgt_interface       => "eth0",
mgt_ip              => "192.168.200.100",
mgt_gateway         => "192.168.200.1",
dns_servers         => "192.168.220.254",
dns_search          => "dmz-pod2.lab",
 }
}

OPTIONAL: When using the networking::interfaces class, you are no longer required to define ip and domain within cobbler::node of the cobbler-nodes.pp file. Here is an example cobbler::node node definition:

cobbler::node { "<compute_node>":
  mac => "A4:4C:11:13:11:AA",
  profile => "precise-x86_64-auto",
  preseed => "/etc/cobbler/preseeds/cisco-preseed",
  power_address => "192.168.254.6",
  power_type => "ipmitool",
  power_user => "admin",
  power_password => "password",
}

To use the networking puppet module:

cd ./puppet/modules
git clone https://github.com/danehans/puppet-networking.git networking

Additional documentation is coming.