2016-06-25

This post is about containers, a construct similar to virtual machines (VM) but so much lightweight that you can easily create a dozen on your desktop Ubuntu!

A VM virtualizes a whole computer and then you install in there the guest operating system. In contrast, a container reuses the host Linux kernel and simply contains just the root filesystem (aka runtimes) of our choice. The Linux kernel has several features that rigidly separate the running Linux container from our host computer (i.e. our desktop Ubuntu).

By themselves, Linux containers would need some manual work to manage them directly. Fortunately, there is LXD (pronounced Lex-deeh), a service that manages Linux containers for us.

We will see how to

setup our Ubuntu desktop for containers,

create a container,

install a Web server,

test it a bit, and

clear everything up.

Set up your Ubuntu for containers

If you have Ubuntu 16.04, then you are ready to go. Just install a couple of extra packages that we see below. If you have Ubuntu 14.04.x or Ubuntu 15.10, see LXD 2.0: Installing and configuring LXD [2/12] for some extra steps, then come back.

Make sure the package list is up-to-date:

Install the lxd package:

If you have Ubuntu 16.04, you can enable the feature to store your container files in a ZFS filesystem. The Linux kernel in Ubuntu 16.04 includes the necessary kernel modules for ZFS. For LXD to use ZFS for storage, we just need to install a package with ZFS utilities. Without ZFS, the containers would be stored as separate files on the host filesystem. With ZFS, we have features like copy-on-write which makes the tasks much faster.

Install the zfsutils-linux package (if you have Ubuntu 16.04.x):

Once you installed the LXD package on the desktop Ubuntu, the package installation scripts should have added you to the lxd group. If your desktop account is a member of that group, then your account can manage containers with LXD and can avoid adding sudo in front of all commands. The way Linux works, you would need to log out from the desktop session and then log in again to activate the lxd group membership. (If you are an advanced user, you can avoid the re-login by newgrp lxd in your current shell).

Before use, LXD should be initialized with our storage choice and networking choice.

Initialize lxd for storage and networking by running the following command:

We created the ZFS pool as a filesystem inside a (single) file, not a block device (i.e. in a partition), thus no need for extra partitioning. In the example I specified 30GB, and this space will come from the root (/) filesystem. If you want to look at this file, it is at /var/lib/lxd/zfs.img.

That’s it! The initial configuration has been completed. For troubleshooting or background information, see https://www.stgraber.org/2016/03/15/lxd-2-0-installing-and-configuring-lxd-212/

Create your first container

All management commands with LXD are available through the lxc command. We run lxc with some parameters and that’s how we manage containers.

to get a list of installed containers. Obviously, the list will be empty but it verifies that all are fine.

shows the list of (cached) images that we can use to launch a container. Obviously, the list will be empty but it verifies that all are fine.

shows the list of available remote images that we can use to download and launch as containers. This specific list shows Ubuntu images.

shows the list of available remote images for various distributions that we can use to download and launch as containers. This specific list shows all sort of distributions like Alpine, Debian, Gentoo, Opensuse and Fedora.

Let’s launch a container with Ubuntu 16.04 and call it c1:

We used the launch action, then selected the image ubuntu:x (x is an alias for the Xenial/16.04 image) and lastly we use the name c1 for our container.

Let’s view our first installed container,

Our first container c1 is running and it has an IP address (accessible locally). It is ready to be used!

Install a Web server

We can run commands in our container. The action for running commands, is exec.

After the action exec, we specify the container and finally we type command to run inside the container. The uptime is just 2 minutes, it’s a fresh container :-).

The — thing on the command line has to do with parameter processing of our shell. If our command does not have any parameters, we can safely omit the –.

This is an example that requires the –, because for our command we use the parameter -h. If you omit the –, you get an error.

Let’s get a shell in the container, and update the package list already.

We are going to install nginx as our Web server. nginx is somewhat cooler than Apache Web server.

Let’s view our Web server with our browser. Remeber the IP address you got 10.173.82.158, so I enter it into my browser.



Let’s make a small change in the text of that page. Back inside our container, we enter the directory with the default HTML page.

We can edit the file with nano, then save



Finally, let’s check the page again,



Clearing up

Let’s clear up the container by deleting it. We can easily create new ones when we need them.

We stopped (shutdown) the container, then we deleted it.

That’s all. There are many more ideas on what do with containers. Here are the first steps on setting up our Ubuntu desktop and trying out one such container.

Show more