Author: Jacqueline

Docker for impatient newbies part 1: Getting started

If you are one of the few that have not yet taken the plunge into Docker don’t feel ashamed. I haven’t either. I’ve been too busy doing other stuff lately. But yesterday I actually started using Docker because I had a real life use case. I have a Node application that I wanted to ship in a Docker container. And I wanted it quickly.

So here is what you need to package a Node application in a Docker container.

Docker for Mac and Docker for Windows

Step 1 is to install the Docker engine for Mac or Windows. Head over to and install Docker for Mac or Docker for Windows.

What happens? You just transformed your machine into a Docker engine. With this Docker engine you can host containers, from images containing your applications.
This is how you get started and start developing. Production environments are better installed in another Docker Engine, somewhere in the cloud or in a datacenter.

Docker on Linux

The Docker Engine on Linux is in its natural habitat and installing it (on Ubuntu Xenial) is as easy as:

sudo apt-key adv --keyserver hkp:// --recv-keys 58118E89F3A912897C070ADBF76221572C52609D
echo "deb ubuntu-xenial main" | sudo tee /etc/apt/sources.list.d/docker.list
sudo apt-get update
apt-cache policy docker-engine
sudo apt-get install -y docker-engine

sudo usermod -aG docker $USER

First steps

First, check the installation:

docker version

1.12 is the latest version:

Let's go ahead and create our first image for a Node Express app.

Package a Node.js Express app

Cd into a Node application you want to package, or clone mine. If you clone mine, install Node.js on your platform if you haven't already. and check if it runs.

git clone
cd node-express-starter
npm install
gulp serve

If all is well you will be presented with a static website. The theme is from Black Tie.

First step is to create a Dockerfile.
Assuming you are still in the node-express-starter dir, run:

touch Dockerfile

This will be the folder structure from node-express-starter:

├── Dockerfile
├── app
├── bower.json
├── gulpfile.js
├── main.js
├── node_modules
└── package.json

This is how the Dockerfile should look like:

FROM mhart/alpine-node:6.3.0


# copy all the files from the Node app to /var/www in the container:
COPY  . /var/www

#set it as workdir
WORKDIR /var/www

#install dependencies from package.json
RUN npm install

# Expose port 3000 from the container to the host

ENTRYPOINT ["node","app.js"]

What does this mean?

  • FROM: here we need to put the base image for our image. Lots of these images yoy can find on the [Docker Hub]( I want the image to be as small as possible. So that's why I'm using an [Alpine]( based image.
  • COPY: I copy all the files from my node project to the /var/www folder on my Docker image
  • WORKDIR: /var/www is the workdir (pwd)
  • RUN: while in /var/www, the npm install is run to install all dependencies of the Node app.
  • EXPOSE: here you can set the port for communication with the outside world. My Node app runs on port 3000
  • ENTRYPOINT: the command 'node app.js' will run upon starting the container (not when creating the image of course)

Now create the container image. You should still be in the node-express-starter dir. Run the following command (don't forget the dot):

docker build -t jacqueline/impatient .

This has created a Docker image with the tag (-t) 'jacqueline/impatient'.

Now run it

docker run -d -p 8080:3000 jacqueline/impatient

With -d the container instance is ran as a daemon and the -p is the port redirection (3000 we exposed in our image will be forwarded to port 8080)

Now head over to http://localhost :8080 and you should be presented with the Node web application:


  • We installed the Docker Engine on our computer
  • We created a Docker image containing our Node app
  • We started an instance of this image


Stopping and cleaning up

If you work with containers, lots of harddrive space will be consumed. This is how you clean up stuff.

First, check which containers are currently running with 'docker ps':

docker ps

Obviously, the one we just ran is active:
Notice the container ID. You can stop this instance with:

docker stop 45

Next, delete the container. First list all the containers:

docker ps -a

Again, notice the ID:

docker rm 45

Now that the container is deleted, we can go ahead and remove the image. First check which images are present:

docker images

There are 2. I'm only going to remove my own image, not the Alpine one.

docker rmi 78

In the screenshot above notice how I forgot to delete the container, but I could still delete the image with the -f flag.

The End. But will be continued.

Of course now that we took the Docker plunge, we now want to host our container in the cloud. At Digital Ocean, AWS or even Azure. This will be covered in part 2.

Build a C# REST API and consume it with Powershell (and write Pester tests!) (part 2)

This is the sequel to part 1.
Now we are going to query the AdventureWorks Database.

Install the AdventureWorks database

The script below downloads the AdventureWorks database for SQL Server 2014, extracts it in c:\temp and restores it. As you can see I had to change the data and the log locations because the original AdventureWorksDb is created in another version of SQL Server.

With the SQL Powershell commandlets you have to create 2 'Microsoft.SqlServer.Management.Smo.RelocateFile' objects to do so. Now, if you have both SQL Server 2014 Express and Visual Studio 2015 Community Edition installed, the SQL Server Management dlls get messed up, because both version 12 and 13 are loaded in the app domain. You can check that with

[appdomain]::CurrentDomain.GetAssemblies() | ? { $_.Location -like "*smo*" } 

This knowledge results in this script. See also this question from StackOverflow.

New-Item 'c:\temp' -ItemType Directory -Force 
Set-Location C:\temp
choco install 7zip.commandline --yes --force
wget -OutFile
7z e .\

Import-Module SQLPS

$RelocateData = New-Object 'Microsoft.SqlServer.Management.Smo.RelocateFile, Microsoft.SqlServer.SmoExtended, Version=, Culture=neutral, PublicKeyToken=89845dcd8080cc91' -ArgumentList "AdventureWorks2014_Data", "C:\Program Files\Microsoft SQL Server\MSSQL12.SQLEXPRESS\MSSQL\DATA\AdventureWorks2014.mdf"

$RelocateLog = New-Object 'Microsoft.SqlServer.Management.Smo.RelocateFile, Microsoft.SqlServer.SmoExtended, Version=, Culture=neutral, PublicKeyToken=89845dcd8080cc91' -ArgumentList "AdventureWorks2014_Log", "C:\Program Files\Microsoft SQL Server\MSSQL12.SQLEXPRESS\MSSQL\Log\AdventureWorks2014.ldf"

#Invoke-Sqlcmd -ServerInstance '.\sqlexpress' -Query "DROP DATABASE AdventureWorks2014"
Restore-SqlDatabase -ServerInstance localhost\sqlexpress -BackupFile C:\temp\AdventureWorks2014.bak -Database AdventureWorks2014 -RelocateFile @($RelocateData,$RelocateLog)

Add Linq to SQL classes

Now it's time to add an Object Relation Mapper. If the app is using Microsoft SQL and is relatively easy (only a few tables, not too much relationships) then why not go ahead and use good old Linq2SQL. It's incredibly easy.

So type CTRL+Shift+A
Add Linq to SQL Classes and call it AdventureWorks.dbml.


Next, open Server Explorer and drag the Person table to the canvas like this:


Now build the Project. And we are done. This simple drag and drop action created a Person class and added the ConnectionString to the Web.Config file.

Create a Person Controller

In the Controllers folder, add a new Class named PersonController.

Change the first Get method as follows:

// GET: api/Person
public IQueryable<Person> Get()
       var db = new AdventureWorksDataContext();
       var people = db.Persons.Take(10);
       return people;

So here we changed the return type in 'Queryable', because that is the return type of a AdventureWorksDataContext collection. You see that we only pick 10 records because the Person table has almost 20.000 records and it takes a long time to load them all.

Now, hit build and start debugging. Fire up Powershell and type:

curl http://localhost:49491/api/person

You may need to change the url to match yours. The result should be:


Consume the api with Powershell and test with Pester

A tip: read because it explains in a very concise way how to use Pester. It's great to be able to write unit tests for Powershell. If you are not sure why you should be writing tests, read this.

Shouldn't I test the api as well? Yes, I definitely should and I will, but it's not in scope of this article. So let's get started with Powershell.

First, add a new Test:

New-Fixture -Name Get-AWPeople -Path .\AWcmdlets -Verbose

This creates 2 files in the AWcmdlets folder:


This is the contents of the Get-AWPeople.Tests.ps1 file:

So let's invoke a test:


Of course, it fails. Now, let's write code in the Get-AWPeople.ps1 file:

function Get-AWPeople 
    param (
       $data = ((Invoke-WebRequest $Uri).content) | ` 
       Write-Output $data.count


We would expect the output to be 10 of course, because our api returns only 10 entries.

Next, write the test:

$here = Split-Path -Parent $MyInvocation.MyCommand.Path
$sut = (Split-Path -Leaf $MyInvocation.MyCommand.Path).Replace(".Tests.", ".")
. "$here\$sut"

Describe "Get-AWPeople" {
    It "returns 10 entries from the Person table" {
        $result = Get-AWPeople -Uri "http://localhost:49491/api/person"
        $result | Should Be 0

Let's run the test again:


Now change the 0 in the file to 10 and run Invoke-Pester:


And we have a success. Now on to write some more functionality!


Query a database through a C# REST API with Powershell (part 1)

It is probably known that you can query an SQL database in Powershell relatively easy, but wouldn't it be great to quickly write a REST API in front of the database? So that you can add business logic if you wish? And use Powershell as a REST client? And then be able to code a decent frontend for the API for whatever device?

Let's get started!
In this series I will first create a WebApi from scratch. Of course, you can also use the templates in Visual Studio, but I prefer to have a bit of knowledge of the code that's in my project. It's not that hard and you will end up with a clean code base.

Step 1. Get your dev environment ready

You can use a Vagrant box. If you use this Vagrantfile a install.ps1 script will be copied to your desktop. Run it, grab a coffee or go shopping because we are on Windows and Windows apps can be huge.

Step 2. Getting the VS Project in place

Start Visual Studio
Create a new empty solution:


I named the empty solution BusinessApp (I'm lacking inspiration for a better name).

Then right click the newly made solution in the Solution Explorer (the pane on the right) and click Add and the New Project:







I named the new Project BusinessApp.Api. If you set your solution up like this you can add more projects as you continue extending the app, for example for an Angular (or whatever framework) frontend, or if you want to separate your datalayer. You can also put your Powershell client modules in a separate project if you wish.

Then open up the Nuget Package Manager Console and install the WebApi dll's:

Install-Package Microsoft.AspNet.WebApi

Make sure to choose the correct Package source (Microsoft and .NET).

Step 3. Add routing

Add a new folder and name it App_Start.
Create a new class in the folder and name it WebApiConfig.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Web.Http;

namespace BusinessApp.Api
    public static class WebApiConfig
        public static void Register(HttpConfiguration config)
            GlobalConfiguration.Configuration.Formatters.JsonFormatter.SerializerSettings.ReferenceLoopHandling = Newtonsoft.Json.ReferenceLoopHandling.Ignore;

                name: "DefaultApi",
                routeTemplate: "api/{controller}/{id}",
                defaults: new { id = RouteParameter.Optional }

In this class we configure that we want our api to return and consume json. Also, we configure our routes to match the controller name, followed by id, wich is optional. E.g would match a controllername Employees, and it would return employee with id 1.

Step 4. Enable CORS

We need to enable CORS else we won't be able to consume the api from from another domain outside the domain from which the resource originated. In a production web environment you should configure this very carefully. I will CORS very permissive because I want my code to work.

Install CORS with in Nuget console:

Install-Package Microsoft.AspNet.WebApi.Cors

Then modify the WebApiConfig.cs class as follows:

using System.Web;
using System.Web.Http;
using System.Web.Http.Cors;

namespace BusinessApp.Api
    public static class WebApiConfig
        public static void Register(HttpConfiguration config)
            var cors = new EnableCorsAttribute("*", "*", "*");
            GlobalConfiguration.Configuration.Formatters.JsonFormatter.SerializerSettings.ReferenceLoopHandling = Newtonsoft.Json.ReferenceLoopHandling.Ignore;

Step 5. Add a Controller

  • Create a folder named 'Controllers'
  • Right click the Controllers folder and click Add and then Controller
  • Click Web API 2 Controller with read/write actions.


I named the Controller Test Controller.

Step 5. Add a Global.asax file

We need to add a Global.asax file to call the WebApiConfig.cs methods at startup.

Right click the solution, click Add, click New Item and search for Global.asax, then Add it.


Modify Global.asax (see the highlighted lines):

using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;
using System.Web.Http;
using System.Web.Security;
using System.Web.SessionState;

namespace BusinessApp.Api
    public class Global : System.Web.HttpApplication

        protected void Application_Start(object sender, EventArgs e)


Step 6. Test the API

Hit F5 and browse to http://localhost:/api/test


And it works. You can also consume the API with Powershell at this point:

((Invoke-WebRequest http://localhost:53601/api/test).content) | ConvertFrom-Json

It should return value1 and value2.

Done! Now let's query a database. This will be explained in Part 2.

Building a development environment with Vagrant, Packer, Windows 10 and Server 2016

Here is another post about creating Windows demo environments with Vagrant. This time we create our own Windows 10 LTSB and Windows 2016 CTP 5 core boxes from scratch with Packer.


You can checkout the Packer templates on my Github page here:


For the impatient: I published my boxes at Hashicorp here. You can use them right away with Vagrant and Virtualbox. I'm currently writing Parallels providers as well.

Here is how.
Install Virtualbox and Vagrant for your OS.
Create a folderstructure on your machine for the Vagrant boxes. Then issue a 'vagrant init':

#Create a folder structure to host the Vagrant boxes 
#Can be named anything and placed anywhere you like
New-Item \Boxes\Win10 -Type Directory -Force; cd C:\Boxes\Win10

vagrant init jacqinthebox/windows10LTSB
vagrant up --provider Virtualbox

And now we wait!
When the box has finished downloading you can provision the box to your liking and sysprep it (there is an autounnattend.xml already in c:\logs).
I currently use this Vagrantfile:

# -*- mode: ruby -*-
# vi: set ft=ruby :

$sysprep = <<'SYSPREP'
& $env:windir\system32\sysprep\sysprep.exe /generalize /oobe /unattend:C:\logs\unattend.xml /quiet /shutdown

Vagrant.configure(2) do |config|
  config.vm.define "client01" = "jacqinthebox/windows10LTSB"
  config.vm.hostname = "client01" "private_network", ip: "" :forwarded_port, guest: 3389, host: 33989, id: "rdp", auto_correct: true
config.vm.provision :shell, inline: $sysprep


The box comes with Chocolatey and Package Management preinstalled, so you can install Chrome with a simple:

Install-Package -Providername chocolatey googlechrome -ForceBootstrap -Force

OK, onto how the box was built and why.

Why build a Windows demo environment?

Indeed, why on earth would I want to build a demo environment? Well, I am a Powershell trainer. And I do most of my trainings at the customer premises. So there's no training equipment and I always ask my students to bring their own device. Now how would I make sure we are working on the same machines and the demo's work? Enter Vagrant.

Why not just create sysprepped golden images and clone them all the time?

I love Vagrant. I love automation! Here are some of my other Vagrant posts:

I did learn a lot in the meantime, so I will basically repeat everything in this post. And now Packer supports WinRM, so installing SSH on Windows is no longer necessary.

How to build the box with Packer

Install Vagrant, Virtualbox and Packer. I'm on Windows 10 and using OneGet, indeed, from the commandline! Hurray!

Install-Package -ProviderName Chocolatey - ForceBootstrap -Force vagrant,virtualbox,packer

Then clone the packer-templates Git repo:

git clone

Now build the box like this and add it to Vagrant:

packer build -only virtualbox-iso windows_2016.json
vagrant box add --name windows_2016

How does Packer work?

Basically Packer builds the VM with parameters it reads from a json file.
This is the json file for the Windows 10 LTSB machine:

  "builders": [
      "type": "virtualbox-iso",
      "iso_url": "{{user `iso_url`}}",
      "iso_checksum_type": "{{user `iso_checksum_type`}}",
      "iso_checksum": "{{user `iso_checksum`}}",
      "headless": false,
      "guest_additions_mode": "attach",
      "boot_wait": "2m",
      "communicator": "winrm",
      "winrm_username": "vagrant",
      "winrm_password": "vagrant",
      "winrm_timeout": "5h",
      "shutdown_command": "shutdown /s /t 10 /f /d p:4:1 /c \"Packer Shutdown\"",
      "shutdown_timeout": "15m",
      "guest_os_type": "Windows81_64",
      "disk_size": 61440,
      "floppy_files": [
        "{{user `autounattend`}}",
      "vboxmanage": [
   "provisioners": [
      "type": "powershell",
      "scripts": [
  "post-processors": [
      "type": "vagrant",
      "keep_input_artifact": false,
      "output": "windows_10_{{.Provider}}.box",
      "vagrantfile_template": "vagrantfile-windows_10.template"
  "variables": {
    "iso_url": "",
    "iso_checksum_type": "md5",
    "iso_checksum": "c22bc85b93eb7cc59193f12f30538f78",
     "autounattend": "./answer_files/10/Autounattend.xml"

So, the whole creation process of a Vagrant box goes something like this:

  • Packer creates and configures the VM
  • Packer attaches the floppy files to the VM
  • It starts installing the OS, the Windows Installer grabs the answer file from the floppy
  • In the answer file the Vagrant user is created
  • In the answer file we will call a powershell script (bootstrap.ps1) which enables WinRM
  • As soon as WinRM is available, Packer will start provisioning scripts the box by executing provision.ps1
  • By running provision.ps1 the guest additions are installed and the box is compacted with sdelete.exe

Let's dissect parts of this the json file.

 "type": "virtualbox-iso",
 "iso_url": "{{user `iso_url`}}",
 "iso_checksum_type": "{{user `iso_checksum_type`}}",
 "iso_checksum": "{{user `iso_checksum`}}",

Here we declare that we are creating a Virtualbox vm.
The iso_url , iso_checksum_type and iso_checksum are variables, which are set in the variables block in the bottom of the file.

"headless": false,
"guest_additions_mode": "attach",
"boot_wait": "2m",
"communicator": "winrm",
"winrm_username": "vagrant",
"winrm_password": "vagrant",
"winrm_timeout": "5h"

We set headless to false to see what is going on. The guest_additions_mode is attach. Valid options are "upload", "attach", or "disable", but we choose attach because uploading the tools to the VM is not working via WinRM (it does work with SSH, but we are not using SSH). So we attach the guest additions iso and install the tools from there in the provision stage (more about that later). We set the winrm_timeout to 5h. This is the time Packer waits until WinRM becomes available.

"floppy_files": [
     "{{user `autounattend`}}",

This is the part where we define the scripts that are placed on the floppy disk that gets attached to the box.

  "provisioners": [
      "type": "powershell",
      "scripts": [

This part is executed after WinRM is available. In the provison.ps1 script we enable RDP, install chocolatey, install the guest additions and last but not least the disk gets compacted with sdelete.exe.

So there it is! Clone the repo, check out the answer files and the scripts and build your boxes already!

Special thanks goes to..

This article: and this Github repo:

Powershell Profile

Here is a quick and dirty way to create a Powershell profile.

First, start the Powershell terminal and type:

new-item $profile -Force
notepad $profile

An empty profile file appears, and you can type the commands you want to run at start up:

cd ~/Documents

Fortune cookie

If you want to be greeted with a fortune cookie, add the following line:

$fc = ((Invoke-WebRequest | ConvertFrom-Json
write-host `t `"$($fc.fortune)`"

Thanks to the Google Developers Group Nort Colorado for hosting the api!

Some nice ASCII art

Finally, add a nice ASCII picture to spice up your day. I found one here. Clone the repo and save Get-MOTD.ps1 in your $env:HOMEPATH\Documents\WindowsPowershell\ directory.

Your $profile file will look like this:

. $env:HOMEPATH\Documents\WindowsPowershell\Get-MOTD.ps1
write-host ""
write-host ""
$fc = ((Invoke-WebRequest | ConvertFrom-Json
write-host `t `"$($fc.fortune)`"
write-host ""
cd ~

And it if you start the Powershell terminal you will be greeted like this:

2016-04-25 12_57_37-Windows PowerShell

Tip: don't do this in ISE. Also, install posh-git and follow the profile adjustments described on their github site.