LXD/LXC creates lighter systems than VM hypervisors, and it is an easy way to experiment with different software without messing up with your system. You can also use it in production, to run Linux systems with much greater density than on VM hypervisors. The only drawback is that the container must be the same OS as the host (Linux in this case) and it depends on the host for kernel functionality (e.g. modules etc).

In this guide we will build an LXD container with two virtual NICs which are attached to the physical NIC of it’s host using bridged networking.

Prerequisites

• An Ubuntu 16.04 host with two physical NICs

Prepare host networking

Install dependencies

# apt -y install bridge-utils

Setup networking on the host

There are two interfaces on the host enp3s0 and enp5s0. The first one, enp3s0 is the primary and connects to the main, DHCP enabled, internal network. The second one is connected to an isolated internal network where some critical services are running. There is no DHCP in the isolated network.

Prepare your /etc/network/interfaces configuration file like this:

# The loopback network interface 
auto lo
iface lo inet loopback

# The main Bridge
auto br0
iface br0 inet dhcp
    bridge-ifaces enp3s0
    bridge-ports enp3s0
    up ip link set enp3s0 up

# The primary network interface
iface enp3s0 inet manual

# The secondary Bridge
auto br1
iface br1 inet manual
        bridge-ifaces enp5s0
        bridge-ports enp5s0
        up ip link set enp5s0 up

# The secondary network interface
iface enp5s0 inet manual

Setup LXC/ LXD

Installation of LXD

# apt -y install lxd

Prepare a new profile

We need a profile with two NICs. First create a new profile:

# lxc profile copy default dualnic

Then edit the new profile with lxc profile edit dualnic:

config:
  user.network_mode: link-local
description: Dual nic profile
devices:
  eth0:
    name: eth0
    nictype: bridged
    parent: br0
    type: nic
  eth1:
    name: eth1
    nictype: bridged
    parent: br1
    type: nic
name: dualnic

The names of the virtual interfaces of the container will be eth0 for the primary and eth1 for the secondary.

Launch the container

Now we want to launch a Debian jessie container, which we will retrieve from the prepared images on the

# lxc launch images:debian/jessie mycontainer -p dualnic

List the container:

# lxc list
+-------------+---------+----------------------+------+------------+-----------+
|    NAME     |  STATE  |         IPV4         | IPV6 |    TYPE    | SNAPSHOTS |
+-------------+---------+----------------------+------+------------+-----------+
| mycontainer | RUNNING | 192.168.0.122 (eth0) |      | PERSISTENT | 0         |
+-------------+---------+----------------------+------+------------+-----------+

The IP listed above was set by the DHCP server in the primary network.

Connect to the container:

# lxc exec mycontainer bash

Setup container’s networking

Edit /etc/network/interfaces inside our container:

auto lo
iface lo inet loopback

auto eth0
iface eth0 inet dhcp

auto eth1
iface eth1 inet static
        address 172.16.16.100
        netmask 255.255.255.0
        network 172.16.16.0
        broadcast 172.16.16.255
        up ip route add 10.10.10.0/24  via 172.16.16.1 dev eth1
        up ip route add 10.10.20.0/24  via 172.16.16.1 dev eth1

The routes 10.10.10.0/24 and 10.10.20.0/24 are internal routes that need to be reached though the secondary network.

Take a clean snapshot

Before you start working on your container, it is a good idea to get a clean snapshot, so you can revert to it, in case you mess something up.

# lxc snapshot mycontainer

If you want to learn more about the and features I urge you to read the excellent introduction to LXD 2.0 by Stéphane Graber

References

• https://linuxcontainers.org/
• https://www.ubuntu.com/containers/lxd
• https://insights.ubuntu.com/2016/03/14/the-lxd-2-0-story-prologue/
• https://stgraber.org/2016/03/11/lxd-2-0-blog-post-series-012/
• https://insights.ubuntu.com/2015/11/10/converting-eth0-to-br0-and-getting-all-your-lxc-or-lxd-onto-your-lan/