First steps with provisioning of Docker containers using Vagrant as provider image70

First steps with provisioning of Docker containers using Vagrant as provider

Note: the next set of (more advanced) steps with Vagrant and Docker are discussed in this article: Vagrant and Docker – Next and Advanced steps with folders, ports, volumes, linking and more.

Automated environment provisioning and use of virtualization isolate environments is a topic that has become quite important to me – and to many others. From virtual machine technology, such as VMWare, VirtualBox and Oracle VM, to more recently the use of container technology and especially Docker, I am constantly trying to wrap my head around things and find ways of efficiently, smoothly hence automatedly working with environments. Vagrant has been a great tool for me to produce VirtualBox machine images that I use for a host of things including development environments. I have frequently used Puppet in conjunction with Vagrant to do the detailed configuration of the Virtual Machine – including the installation of various software packages.

Docker is of interest for several reasons: Docker containers while isolated share their host operating systems and its physical resources which means that a collection of Docker containers can provide (almost) the same isolation as a series of Virtual Machines but with much smaller overhead – in terms of usages of disk, CPU and memory and in terms of time required for starting up and shutting down. Additionally, the management of Docker containers – from 10Ks of images to start from and a layered mechanism to extend and revert – has a number of attractive options. A Docker container – once configured and runnable – can be distributed fairly easily – using a public or private registry – and can also be deployed to a growing number of enterprise stacks (such as OpenStack cluster) and public cloud providers.

Docker requires a Linux host as Docker itself leverages the LXC (Linux Containers) mechanism in Linux. This means that in order to work with Docker on non-Linux systems – Windows, Mac OS X, Solaris – we first need to set up a Virtual Machine running Linux. The special boot2docker package is a lightweight Linux distribution made specifically to run Docker containers. A quick way on Windows and OS X to get started with Docker is by using boot2docker as an intermediate mechanism.

The association between Vagrant and Docker is threefold:

  1. automate the building of Docker containers
  2. coordinate running of Docker containers
  3. provide a Docker enabled host VM on which Docker is run

image

1. Vagrant can be used to run Docker containers – with vagrant up resulting in the configured container(s) being run. Vagrant supports special commands to produce logging from Docker containers and to run one-off commands in a container. The Vagrant way of defining folder mapping is leveraged through Docker volume instructions, resulting in host folders being accessible inside the Docker container.

2. Vagrant has embraced Docker as a provider, just as it supports providers for VirtualBox and VMWare. This means that a Vagrant configuration file can describe a Docker container just as it can describe the configuration of a VirtualBox VM. When Vagrant runs with a configuration for Docker, it will start provisioning the Docker container as specified. This means that anyone comfortable using Vagrant as mechanism for describing and provisioning environments can quickly apply their existing knowledge and skills to Docker containers instead for classic VM images. The port forwarding and host-to-guest folder mapping that we configure in the Vagrant file is now applied to host-to-Docker container interaction.

3. One very nice additional touch is that Vagrant is aware of the fact that Docker containers cannot run natively at present on Windows or Mac OS X. When Vagrant is asked to provision a Docker container on one of these operating systems, it can either automatically engage boot2docker as a vehicle to create and run the Docker container in or provision a Linux based VM image that it then enables for Docker and creates the Docker container into. Using a traditional Vagrant file, we can describe the Docker host VM that we want Vagrant to create or leverage (if it already exists). This host VM can of course be made to look much more like a production Linux host than the boot2docker environment. The same Docker host VM that Vagrant creates can be reused for many Docker containers provided by Vagrant: a single Linux VM suffices for running multiple containers that can mutually be linked and share file system folders. Vagrant can also provision such a Docker VM when it is running on Linux on which it could also natively install Docker can build containers.

This article is an overview of my first steps with the combination of Vagrant and Docker (on Windows) – and really not much more than a rehash of the best parts provided in several great articles listed in the Resources section. It is a stepping stone towards automated environment provisioning using Vagrant, Docker and Puppet and provides some examples that address the main challenges I faced in getting started.

 

Simple first step – build and run a Docker container

Using three files on my Windows laptop that has both VirtualBox and Vagrant installedimage

I will create and run a very simple Docker container, based on the Ubuntu 14.04 image. The container will do very little of interest: it pings localhost 51 times and exits again. This gives us some time to inspect the container, ping it, attach to it and look at its output. Later on in this article, things will get a little more interesting.

image

The contents of the Vagrantfile:

ENV['VAGRANT_DEFAULT_PROVIDER'] = 'docker'

Vagrant.configure("2") do |config|

config.vm.define "my-little-container" do |m|

m.vm.provider :docker do |d|
d.name = 'my-little-container'
d.build_dir = "."
d.cmd = ["ping", "-c 51", "127.0.0.1"]
d.remains_running = true
d.vagrant_machine = "dockerhostvm"
d.vagrant_vagrantfile = "./DockerHostVagrantfile"
end
end
end

The DockerHostVagrantFile that it refers to:

Vagrant.configure("2") do |config|

config.vm.provision "docker"

# The following line terminates all ssh connections. Therefore Vagrant will be forced to reconnect.
# That's a workaround to have the docker command in the PATH
# Command: "docker" "ps" "-a" "-q" "--no-trunc"
# without it, I run into this error:
# Stderr: Get http:///var/run/docker.sock/v1.19/containers/json?all=1: dial unix /var/run/docker.sock: permission denied.
# Are you trying to connect to a TLS-enabled daemon without TLS?

config.vm.provision "shell", inline:
"ps aux | grep 'sshd:' | awk '{print $2}' | xargs kill"

config.vm.define "dockerhostvm"
config.vm.box = "ubuntu/trusty64"

config.vm.provider :virtualbox do |vb|
vb.name = "dockerhostvm"
end

end

Finally the Docker file:

FROM ubuntu:14.04

RUN mkdir /u01 && \
chmod a+xr /u01
COPY /files/readme.txt /u01/

The command to get things started is vagrant up. This will have vagrant start provisioning, based on the Vagrantfile. Vagrant will realize that we ask for Docker and that the operating system is Windows that does not (yet) natively support Docker. Therefore, a Docker enabled host VM is required. Instead of creating such a VM based on boot2docker – which is default behavior – Vagrant will interpret the reference to the DockerHostVagrantFile. This file contains a straightforward definition for a VirtualBox VM based on an Ubuntu base box (ubuntu/trusty64). The file can also specify a private network configuration with a preset IP address that we can later on use to connect from the Windows host into the Docker Host (VM). We could also specify other VM settings – such as CPUs and memory – just as we can do for any VM created by Vagrant.

The Vagrant logging shows the creation of the dockerhostvm Virtual Box VM:

image

Once the installation of the Virtual Box Guest Additions on this VM is complete, then the default mapping of the Vagrant directory to the VM is created.  Next, the VM is Docker-enabled. A single in line script command is executed on the dockerhostvm to ensure we do not run into a permission denied error (see comment in DockerHostVagrantFile). In order to make the Dockerfile directory available during container build, this directory also needs to be mapped into the dockerhostvm.

image

Then, the provisioning of the Docker container can take place. It is visualized in this next image:

image

First, the Ubuntu image is pulled in :

image

Next the statements in the Dockerfile – such as COPY and RUN are executed. Note that the directory that contains the Dockerfile is available in the build context as the root directory. I have added a files subdirectory to this directory and created a text file called readme.txt in it.

image

The steps in the Dockerfile have Docker create a directory /u01 and copy the file readme.txt from the /files/ directory in the build contxt to this new /u01 target directory.

image

 

Finally the container is created and started.

image

The CMD parameter in the Docker provider section in the Vagrantfile – [“ping”, “-c 51”, “127.0.0.1”] – tells Docker to have the container execute the ping statement as soon as it is running – pinging the localhost (127.0.0.1). This statement is repeated 51 times.

We can see the output from the container on the Windows host using the command vagrant docker-logs:

image

 

This tells us that the container is running and producing output (the output from the ping command).

With vagrant global-status, we can check on the machines that Vagrant controls. The result from our recent activities is that two entries have been added: the dockerhostvm (created by the virtualbox provider) and the default , provided through the docker provider (inside the dockerhostvm – although we cannot tell that fact from this listing).

image

Note that all vagrant commands by default act on the Docker container – listed here as default. Vagrant destroy, halt, up etc. all act on that container and not on the dockerhostvm  Virtual Box VM. If you want vagrant to act on that machine, the commands need to make use of the id of the VM – for example vagrant halt <machine id>.

Using vagrant ssh <machine id for dockerhostvm> we can connect into the Docker host – where we can issue Docker commands, such as docker ps -a (to check on containers):

image

Check on containers running or exited:

image

Start the container, check on its IP address using docker inspect -f ‘{{ .NetworkSettings.IPAddress }}’ <container id> and ping to it to verify its accessibility:

image

Using docker attach <container id> we can attach to a running container using the container’s ID or name, either to view its ongoing output or to control it interactively.

Alternatively, we can start the container in interactive mode (which is more or less a combination of start and attach) using docker start -i  <container id> which in this case means we start viewing the output from the repeated ping command.

image

A running container – executing some kind of job – can be simultaneously accessed to perform other processes. Using docker exec -it <container -id> bash for example, we start an interactive shell in an already running container. So while the my-little-container is doing its 51 laps of pinging, we can do other things as well. Such as … ping too?

This screenshot shows me listing the containers, starting the container (using its container id) and then when it is running also using docker exec to run a command (in this case bash to open a shell) against that same container. Also notice that I use the -it flags with this command, to give an interactive session with terminal.

image

In this session I again verify that the readme.txt has been copied to a newly created directory /u01 inside the container my-little-container. Note that our second session is killed when the initial session (process 1) that ran when the container was started has completed.

Executing commands in a container can also be done from the Vagrant (or Windows) host, using the vagrant instruction docker-run. This will start a container (not reuse an already running one), execute the command and close the container. To open a bash shell into the my-little-container container from the Windows host, use: vagrant docker-run -t  — bash.

image

 

Stopping a running container can be done from the dockerhostvm using regular docker commands. From the Vagrant host (the Windows laptop) we can also stop a container, using vagrant halt.  This next picture shows on the left side the actions on the Windows machine and the status of the Docker containers inspected inside dockerhostvm on the right side. It should be obvious how a vagrant up causes the container the run and a vagrant halt stops the container again.

image

The command vagrant destroy acts on the Docker container – and removes it from the dockerhostvm. It can subsequently be restored with a fresh vagrant up:

image Recreating the container is simple:

image

Note that using the command docker history <image id> we can get an insight in the layers that make up (the images that provides the foundation for) our container – including the layers we added ourselves through the statements in the Dockerfile:

image

 

Summary

This article showed how you can leverage Vagrant to create and manage a simple Docker container as well as a Virtual Box Linux VM that provides the Docker engine & infrastructure. Execution of simple Docker files was demonstrated including how to make files from the Windows or Vagrant host available in the Docker build context. We have seen the commands to attach to the running container and look what is going on. We have also seen how we can execute command against the Docker container from either the dockerhostvm or the Windows/Vagrant host. Some Docker container management and inspection commands where shown – including inspect and history.

In a next article, I will discuss more complex Docker files that create more interesting Docker containers. I will also discuss more advanced options and operations including the installation of Java, mapping folders (from both the Windows host and the dockerhostvm), using port forwarding to access a Java Web Service running in the Vagrant container from the Windows host, linking containers and using a Data Container to prevent having to create temporary files in the Docker layer structure during the container build process

 

Resources

Very good overview: A Not Very Short Introduction to Docker

Setting up a development environment using Docker and Vagrant – by Mario Loriedo on Blog Zenika

Development Environments With Vagrant and Docker – Christian Soronellas – introducing Vagrant Provisioner to provision Docker Containers (with Nginx and PHP-FPM)

Rails development with Docker and Vagrant Pablo Acuña – very clear overview of getting a Rails and MySQL pair of Docker containers created (built) and running using Vagrant

Rapid Local Development with Vagrant, Docker and Node.js – Kev Zettler

Building Containerized Apps With Vagrant – on Willem’s Fizzy Logic

Docker Networking Made Simple or 3 Ways to Connect LXC Containers by Lukas Pustina

Vagrant Documentation – Docker Commands, Basic Usage Docker Provider

A quick introduction to Docker and Using Docker with Vagrant – on Scott’s Weblog

Getting Started with Docker on Servers for Hackers.com

Getting Started with the Nginx Web Server by Travis Reddell

Forward Ports from boot2docker using the Vagrant Docker provider – on StackOverflow

Forwarding ports from a docker container with Vagrant 1.6

Vagrant, Docker & CoreOS by Ashley Clarke

vagrant-docker-exec – plugin to run docker exec commands from your host

SlideShare presentation Linux containers and docker by  Fabio Fumarola

Multi-Service Local Development Environment with Vagrant and Docker by Chris Brantley

My own earlier articles on Docker: My First Steps with Docker – starting from Windows as host and Docker – Take Two – Starting From Windows with Linux VM as Docker Host

One Response

  1. Gaurav Jain December 8, 2016