Azure on Elvenware V

This page covers Windows Azure and related subjects.

Azure is Miscrosoft’s cloud development platform. Like Amazon Web Services and other tools, there are several different services, ranging from storage, to VMs to platforms for installing applications.

Here are links to key areas in the Azure world:

As with Amazon, you need an account and a credit card to start working in this world. There is a three month free trial available. Again like Amazon, you probably already have an account with Microsoft, but you will need to configure it to use the Azure services.

Install Tips

Using the Microsoft Web Platform Installer, make sure you have the latest Windows Azure SDK for .NET and the Visual Studio Tools. This will give you the

  • Azure SDK
  • The Azure Emulator
  • Visual Studio Tools
Azure SDK .NE Install_
Figure 01: Installing the Windows Azure SDK for .NET with the Web Platform Installer

The success screen for the Azure .net
Install |:–:| | Figure 02: The Successful Install of Azure for .NET and VS |

When you are done, you should be able to choose File | New Project ASP.NET Web Application in Visual Studio, open the Project Manager, right click on your projects main node (WebApplication1) and select Add Windows Azure deployment project.

While you have the Web Installer open, there are other tools that you want to install. Two of the most important are the Python Tools for Visual Studio and the Windows Azure SDK for NodeJS. These installs include IIS Node and the Node Package Manager.

Python tools for Visual Studio and Node JS

Figure 03: Python tools for Visual Studio and Windows Azure SDK for Node JS Support

Python Tools for VS and NodeJS for Azure Success

Figure 04: Python Tools for VS and NodeJS for Azure Success Screen

When looking at the Web Platform Installer, you probably noticed that there is an option to install Visual Studio Service Pack 1. In many cases, if you check the about screen for Visual Studio, you will find that you already have SP1 installed, even though this option is available in the Web Platform Installer. My opinion is that you are all set so long as you have Visual Studio SP1 installed. If, however, you feel a strong need to glog up your system with additional useless DLLs and tools, you can select this option in the Web Platform Installer. When the procedure completes, you will be greeted with the following rather depressing list of tools that are now making your system run even slower than it was. When you go back to the Web Platform Installer, you will find, of course, that you are still told that you have not installed VS SP1. Thank you so very much! None of this is meant to discourage you from installing SP1, which is, of course, an essential and very valuable tool.

Visual Studio SP1

Figure 5: Visual Studio SP1 Effluvia.

Install NodeJS

You will also need to install the Windows Azure SDK for NodeJS:

Here is what the Platform Installer looks like if you have successfully installed this product.

Windows Azure SDK for NodeJS_
Figure 06: Windows Azure SDK for NodeJS

Php Install

Installing PHP for Azure is a bit different. The install is on CodePlex, and all you will need to download is a zip file:

There are set of five directories in the zip file:

Azure install for PHP

You should copy this 22 MB of files to some well known location on your hard drive. As always, make sure there are no spaces in the path to your folder. It is all right if you folder itself has a complex name with no spaces, such as: J:\Src\PHPAzure-4.1.0. You will notice that this SDK is made by a company called RealdolMan, and sponsored by Microsoft. As a result, I suppose it has only semi-official status, but so far I have found it be quite professional.

Python Install

The Python install for Azure, on the other hand, is without official sanction from Microsoft. Nevertheless, it does not really differ so greatly in its fundamentals from the PHP install. The simplest way to install the product is to be sure you have a GitHub client of some kind, links are available for that client at the bottom of this document. Then you can paste in the following URL and pull down the files:

Otherwise, you can just go here, and download the zip file:

Azure Install for Python from GitHub using GitGui

For Python, you will also want to be sure you have installed easy-install, pip, flask, virtualenv and rocket. You should also install the Python tools for Windows Azure Storage


There are four languages with top level support on Azure: C#, NodeJS, PHP and Java. I have also found a way to run Python on Azure, though it does not enjoy official support.

C# {#c#}


Start a NodeJs Powershell in Adminstrator mode by right clicking on the following link and choosing Run as administrator:

Start | All Programs | Windows Azure SDK for NodeJs | Windows Azure Powershell for Node JS

Switch to the drive where you store your source code. Assuming you want to place your node js code at the root of the drive you should type

mkdir \nodejs
cd \nodejs
New-AzureService webroles
Add-AzureNodeWebRole WebRole01
cd WebRole01
notepad++ server.js

Here is the code you should place in server.js:

var http = require('http');
var port = process.env.port || 1337;
var server = http.createServer(function (req, res) {
 res.writeHead(200, { 'Content-Type': 'text/html' });
 res.end('<p>Hello HTTP from the JSNode\n</p>');


If you prefer slightly more valid HTML, then write this instead:

var http = require('http');
var port = process.env.port || 1337;
var server = http.createServer(function (req, res) {
 res.writeHead(200, { 'Content-Type': 'text/html' });
 res.write('<!DOCTYPE html>\n');
 res.write('<html>\n');Running Python
 res.write('\t<p>Hello HTTP from the JSNode</p>\n');


Now start the Azure Emulator:

Start-AzureEmulator -launch

The entire session looks a bit like this:

PS C:\> mkdir nodejs
 Directory: C:\
Mode LastWriteTime Length Name
---- ------------- ------ ----
d---- 6/4/2012 3:48 PM nodejs

PS C:\> cd .\nodejs
PS C:\nodejs> New-AzureService webroles

Service has been created at C:\nodejs\webroles

PS C:\nodejs\webroles> Add-AzureNodeWebRole WebRole01

Role has been created at C:\nodejs\webroles\WebRole01. For easy access to Windows Azure services fr
om your application code, install the Windows Azure client library for Node.js by running ‘npm inst
all azure’.

PS C:\nodejs\webroles> cd .\WebRole01
PS C:\nodejs\webroles\WebRole01> notepad++ .\server.js

C:\nodejs\webroles\WebRole01>"c:\Program Files (x86)\Notepad++\notepad++.exe" .\server.js
PS C:\nodejs\webroles\WebRole01> Start-AzureEmulator -launch

Creating local package...
Starting Emulator...
Role is running at
PS C:\nodejs\webroles\WebRole01>;

You can now use the icon on the taskbar to shutdown the emulator. The next step is to publish the service to Azure itself.


Save the settings to your hard drive:

Azure Note Credentials

Now import the settings. Below you can see a call to Import-AzurePublishSettings where the parameter is the name of the settings file that I downloaded. You can delete or store secretly the file after you download it. The second call actually publishes the service under the name Elvenware001.

Import-AzurePublishSettings .\CharlieCred.publishsettings
Publish-AzureService -Location "North Central US" -Name Elvenware001


Start by opening a command prompt at the root of your project and typing this command:

npm install node-uuid

Then create the following program:

var http = require('http');var uuid = require('node-uuid');

var port = process.env.port || 1337;  var server = http.createServer(function (req, res) {   res.writeHead(200, { 'Content-Type': 'text/html' });   res.write('<!DOCTYPE html>\n');   res.write('<html>\n');
 res.write('<head>\n<title>UUID Demo</title>\n</head>\n');
 res.write('<body>\n');   res.write('\t<h1>UUID Generation</h1>\n');
 res.write('\t<p>' + uuid.v4() + '</p>\n');
 res.write('</body>\n');   res.end('</html>');  })

Run the program in node and in the browser you will see something like this:

UUID Generation

I have placed a sample program like the one quoted above in AndElf in the folder Python/AzureRelated.

Debugging Your Server Side Code

You can debug your server side code in Eclipse. Here is how to proceed:

  • Start Eclipse and optionally set up your project to run in it. You don’t actually have to set up the project to run Node from the IDE, but if you want, you can set everything else up.
  • ChoRunning Python on Azureose Help Install New Software
  • Press the Add button
  • Enter this address:
  • Choose to install the Google Chrome Developer Tools
  • Chrome will restart
  • Go to the command prompt and run your program in Node with the Debug switch:
  • node --debug server.js // See Figure D01
  • In Eclipse, switch to the Debug perspective. (**Window Open  
    Perspective Other Debug**)
  • Choose **Run Debug Configurations** from the menu
  • Select Standalone V8 Vm
  • Click the plus button to get a new configuration
  • Call it Server on Port 5858
  • Set the Host to localhost and the port to 5858
  • Press Apply to save your work // See Figure D02
  • Click the Debug button
  • In the project explorer find the source for the file you want to debug. Look carefully at shots D04 and D05. Notice that I’m opening a file with a long name and lots of underscore characters. I picked this file from the Project Explorer or Script Explorer – I believe either one will work.
  • Open the program in a browser: http://localhost:3000 //Figure D03
  • Put a break point on one of your methods, such as request //Figure D04. If you have trouble setting a breakpoint using the tools, go into your source code, write the command debugger;in your source at the place where you want to stop, and then restart your debug session.
  • Do something in the browser that will fire the call to request
  • Check Eclipse: you should be at the breakpoint. // Figure D05
  • When you are done, you should right click on the project and close it. Either that, or make sure you are running node in Debug mode (Figure D01) when you next open Eclipse. If you don’t do one of these two things, you may get a lot of errors when you open Eclipse. If you do get the errors, right click and close the project and restart Eclipse. Yea, I know, it is pretty funky: but it does work if you baby it along some.
Starting node in debug mode
Figure D01: Starting node in debug mode
Setting up the debug Run Configuration_
Figure D02: Setting up the debug run configuration
Running your script in the browser_
Figure D03: Running the program in a browser

Setting the breakpoint in your
source |:–:| | Figure D04: Setting a breakpoint. Click to expand |

If you are not able to set the breakpoint by double clicking or right clicking in the gutter on the far left of the IDE, then you can instead insert a breakpoint with code by inserting the word debugger:

sdb.listDomains(function(error, result, meta) Running Python{
 if( error ) {
 res.send('listDomains failed: '+ error.Message );
 else {

You should enter this breakpoint not in your original source code file, but in the one with the funny long name. You will be asked to make the file writable, and you should choose yes. Then right click, choose V8 Debugging | Push Changes to VM. Do something in the browser to cause the line of code you want to execute to be activated, and your program will stop as expected when it hits the word debugger. You can also put this word in your original source file, but that seems like an even more unwieldy solution.

Figure D05 shows what it looks like when you hit a breakpoint set with the IDE tools. It will look the same if you manually enter the word debugger in the source, except of course there will be an extra line of code where you entered the word that set the breakpoint:

 else {

Hitting a breakpoint in your server side

Figure D05: Hitting a breakpoint in your server side code. Click to expand

| Error you get if you don't have node running in debug_ |

Figure D06: Error you get if you don’t have node running in debug mode. (See Figure D01.)

When looking at Figure D05, note that the Expressions, Breakpoints and Variables windows are all visible. We are inspecting the variable called result in both the Variablesand the Expressions windows.


Working on Linux:

You can install Node like this:

sudo apt-get install nodejs
sudo apt-get install npm

That might not be the most recent nodejs. If that is hte case, then go here:

If you need to do file difs, use Meld.

Running Node from NotePad++ {#runningNodeFromNotePad++}

You can run your Node programs directly from Notepad++:

  • Select **Run Run** from the menu
  • Using quotes around any paths that contain spaces, type in the path to node and a macro that points to your main script file:
  • “C:\Program Files (x86)\nodejs\node.exe” “$(FULL_CURRENT_PATH)”
  • SavRunning Python on Azuree the command under a name like MyProgram. Specify a shortcut like Ctrl + F5
  • Now you can run your program directly from NotePad++. At any rate, its starts Node, and then you can browse to http://localhost:3000 or something similar to use the program.

Notice that NotePad++ starts up a blank command window that runs your instance of Node. This like the way Node starts, and then seems to “hang” when you run it from the command line. When you are done, you can Alt-Tab to this window and press Ctrl-C to close it.



To get started, make sure have installed Flask, Virtual Env, etc:

You should also have the most recent files AndElf Mercurial.

Copy the Python\AzureRelated\AdditionSqlite project to a directory on your system. Open a command prompt and Navigate to that folder and create virtual environment by typing virtualenv venv:

J:\Src\PyDevProjects\AdditionSqlite>virtualenv venv
New python executable in venv\Scripts\python.exe
Installing setuptools................done.done.
InsInstalling pip...................done.

You can see that I have a copy of AdditionSqlite in a directory on my J drive. You may have installed it somewhere else, but the rest of the command shown above should look about the same on your system. If something goes wrong, that probably means that you don’t have virtual environment installed, or else your C:\Python27\Scripts folder is not on your path.

Your virtual environment contains a local copy of your Python installation, just as it will be configured on Windows Azure. That ensures two things:

  • You have Python installed on Azure
  • It is the same version of Python that you are expecting
  • Your copy of Python is configured exactly as you expect

In short, Virtual Environment gives you the chance to configure your machine in quite specific ways without actually changing any of the global settings on your machine.

Start the virtual environment:

(venv) J:\Src\PyDevProjects\AdditionSqlite>

Notice the (venv) inserted before your command prompt. This shows that you are running inside your virtual environment. Now install Flask into the Python installation for your virtual environment:

(venv) J:\Src\PyDevProjects\AdditionSqlite>pip install Flask

Flask is a fairly complex tool that gives you three great tools:

  • A light weight web server called Werkzeug. (Should not conflict with running instances of IIS or Apache.)
  • A template engine called Jinja2 for manipulating HTML
  • Flask, a library for writing Python web applications

Now you should have the virtual environment configured properly and you are ready to run the program by typing python

(venv) J:\Src\PyDevProjects\AdditionSqlite>python
* Running on* Restarting with reloader

Now you are ready to actually run the program. Fire up a browser and navigate to the url provided at your command prompt:

Or, to say the same thing in human readable format:


You are now browsing to the AdditionSqlite program using the Wekzeug web server. The presence of this web server should not conflict with existing running installs of IIS or Apache. Try the following URLs to see how the program works:

The second time you call show you should see a new row has been inserted in the Sqlite database from which the program is pulling data.

When you are done exploring the program, you can press CTRL-C at the command prompt to close the Web Server. To exit the virtual environment, type deactivate at the command prompt. **

Running Python on Azure

NOTICE: _As Microsoft rolls out there new tools for Azure, it is becoming clear that we will have an alternative to the Azure Emulator and the Azure Cloud Services. Instead of using Cloud Services, which is a Platform as a Service model, we can use the Azure virtual machines, which is an Infrastructure as a Service model. Azure virtual machines is still in beta, but it works fairly well, and it is similar to AWS, so it should be easy for you to use, or at least easy for you to understand from a conceptual point of view. You can create an instance of Windows Server or Linux on Azure, and install your program onto that instance. Just as on AWS, you need to take a moment to open the ports to let your program through. On Azure, this means opening endpoints, and you want to open port 9000 for Python and port 3000 for node, though of course you can run your programs on whatever port you want, including port

  1. Because you can run your programs directly in the Azure virtual machine, this means that you are nearly done writing code if you want to use this option rather than Azure Cloud service. Please note: to use Python on Azure you should use Rocket rather than Werkzeug; read the next view paragraphs to see the simple technique for transforming your Werkzueg code into a Rocket code._

So now you have the program, such as it is, running locally. The next step is to try it in the Azure Cloud Services emulator. To begin, find the copy of the Python Tools for Azure that you downloaded earlier. Though you can work directly in the GIT repository, you should probably copy the entire (small) folder to a new location. Navigate to the WorkerRole\app directory in the Python Azure tools and paste in the contents of AdditionSqlist directory, but without the virtual environment folder:

| The app directory just before running on the Azure_ |

Figure P01: The app directory just before running on the Azure emulator. Notice that I have not copied the venv folder.

We do not copy the virtual environment folder because the Python Azure tools will set that up for us autamatically when it is needed. Before we run the program, you should edit You can see this file in Figure P01. It is a very small text file. All you need to do is uncomment the references to Rocket:

import os
from rocket import Rocket
from packme import app import app

if __if __name__ == "__main__":#
Rocket((os.environ.get('ADDRESS', ''), int(os.environ.get('PORT', 9000))), 'wsgi', {'wsgi_app': app}).start()

Rocket is yet another web server, and for some reason Steve Marx wanted to include it in his version of the Python Azure even though Flask already had a web server. Above you can see the way that runserver.pyshould look after you have uncommented the two references to Rocket.

The console for the Azure cloud looks very different in its most recent incarnation. In Figure P00 you can see the Virtual servers, Cloud Service, databases and Storage Accounts that I have set up in the cloud.

The new interface for the Azure

Figure P02: The services you have configured in the cloud appear in a list.

Figure P04: Setting up the endpoints in the Azure

Figure P03: Setting up the endpoints in the Azure cloud

You can use the Remote Desktop Connection tool that ships with Windows to connect and configure your server. You will probably see a button at the bottom edge of one of the Azure screens that will set up your connection automatically. If you need to do it manually, you first need to know the name of the server you connect to. Look at the endpoints panel (Figure P03) to get the port.

When you sign on, you will be prompted for a user name and password. You supplied the password when you created the VM. You need to attach using the Remote Desktop Connection. You should use Adminstrator as the user and provide the password you gave when you created the VM as your credentials. For the domain, you can probably supply almost anything, but the name of the server itself can probably act as the domain. For instance, if I call my server server01, and I access it at, then you can probably sign on as:


| Connect to the Azure cloud with Remote Desktop_ |

Figure P04: Connect to the Azure cloud with Remote Desktop Connection. Note the port number you get from the endpoint screen, seen below.

And then give the password you assigned when creating the VM. If you’ve lost that password you can always delete the VM and start over.

To install programs on the server that you already have on your local machine, you can simple right click on the folder that contains your node application, choose copy, then pick a destination on your F drive for your server, and pick paste. It might make sense to skip big folders like node_modules and then use NPM on the server after installing node from this address:

Below you can see several program set up to run on an Azure server.

Running three web servers on

Figure P05: Running three web servers running three web apps on a beta of Server 2012 in the Azure Cloud. Click to enlarge.

The screenshot shown in Figure P05 is of a remote desktop connection. Notice that there is a port number shown in the caption bar; you need to use that number when setting up the connection. You can see that I have web apps running at port 80, 3000 and 5000. The browser, seen in the background, displays the program that is running at Port 80 in the Rocket web server.

| Looking at the Azure apps running in the browser on my home_ |

Figure P05: Looking at the Azure apps running in the browser on my home machine

In Figure P03 I’ve minimized my connection to the Azure virtual machine and connected to the applications that it is running in my browser. Notice that I am also running a Node web app on a Linux box in the Amazon Web Server cloud.

Please, may my application come through the firewall. I've
already set up the

Figure P05: Please, may my application come through the firewall. I’ve already set up the endpoint!

Just setting up the endpoints (Figure P04) wasn’t enough. I also had to configure the server to let the ports passed the firewall.

Using Virtual Machines

To deploy this code to an Azure virtual machine, do the following:

  • (If you are in the old console view, look for the link at the bottom that references the preview. Click it.)
  • Select New | Virtual machine (You may have to sign up for the beta. You will have an option to choose Server 2012. Why not go for it?)
  • Attach a disk and assign it to drive A (2 GB would be way more than enough)
  • Open up an endpoint either at port 80 or port 9000
  • Using the standards Windows tools to set up your machine to run Python
  • Run your python code as described above, or edit the to use Port 80

I understand, you want more detail, but this should get you started.

Using Steve Marx’s Solution

To get through the next part, you need to think about the structure of Steve Marx’s code. It looks like this:

-- WorkerRole
--- app

As you can see, we have three folders nested inside one another. In the WorkerRole folder there is a copy of run.cmd. There is also a copy in PythonRole. You should edit the one in WorkerRole so that it calls rather than

Here is the file you need to edit:

Finding the right copy of run.cmd_
Figure P02: Finding the right copy of run.cmd

And here is the change you make to the last line of the file:

cd /d "%~dp0"


REM Use this virtual environment.
call scripts\activate

cd app


Be careful not to confuse this copy of run.cmd with the copy found in the root of the Python Azure tools directory.

Before you can run the Azure emulator, you will have to start a SQL Azure prompt. Choose Start | All Programs | Windows Azure SDK for .NET | Windows Azure Command Prompt

Navigate to the root of the Azure tools for Python and run the script called run.cmd. You may be prompted four or five times to sign over to Steve Balmer your:

  • First Born
  • Your house
  • Second Born
  • Parents

Be sure to click OK at each prompt. Now navigate to:


All should be exactly the same as running it at port 5000, only this time you are emulating how it will work on Azure. If it worked before, but does not work when you run in the emulator, consider whether or not you installed Rocket into your main Python installation. (Oddly enough, it appears to be the C:\Python27 version of Python that the emulator uses. But when you install to Azure itself, it will use the virtual environment. Stever Marx’s Python Azure tools will, however, ensure that Rocket is intalled in Azure proper.

Finally, run the Pack command from the root of the Python tools for Azure, then navigate to the Azure console on the web, and follow the prompts to install the files created by your call to Pack. Please note that on Azure you will be able to read from your database, but inserts will fail, presumably because Azure is a readonly environment. We will address that issue later.

Express and Jade

jade –out tmp/ –path views –pretty –obj ‘{title: “Elvenware” }’ {views/index,views/layout}.jade

Storage and SQL Azure

Additional Notes on Node

Learn about NodeJS:

 npm install azure node-uuid DSInit /sqlInstance:.


Templating engines:

This usually means that you don’t have a default file set for your application. Try explicitly naming the file:


403 - Forbidden: Access is denied

<add value=”server.js” />

    <?xml version="1.0" encoding="utf-8"?>
      For more information on how to configure your ASP.NET application, please visit
        <add key="EMULATED" value="true" />
        <modules runAllManagedModulesForAllRequests="false" />
                <add value="server.js" />
        <!-- indicates that the server.js file is a node.js application
        to be handled by the iisnode module -->
          <add name="iisnode" path="server.js" verb="*" modules="iisnode" />
            <clear />
            <rule name="app" enabled="true" patternSyntax="ECMAScript" stopProcessing="true">
                <match url="server\.js.+" negate="true" />
                <conditions logicalGrouping="MatchAll" trackAllCaptures="false" />
                <action type="Rewrite" url="server.js" />
- [Tortoise GIT](
- [](
- [Azure
Copyright © Charlie Calvert)