Installing and Deploying QATrack+ on Windows Server

Note

This guide assumes you have at least a basic level of familiarity with Windows Server, SQL Server Management Studio and the command line.

New Installation

This guide is going to walk you through installing QATrack+ on a Windows Server 2016 server with IIS 10 serving static assets (images, javascript and stylesheets) and acting as a reverse proxy for a CherryPy web server which serves our Django application (QATrack+). SQL Server 2016 will be used as the database.If you are upgrading an existing installation, please see the sections below on upgrading from v0.2.8 or v0.2.9.

Note

This guide assumes you have SQL Server Management Studio (SSMS) and Internet Information Services (IIS) installed/enabled

The steps we will be undertaking are:

Installing git

Go to http://git-scm.com and download the latest version of git (msysgit) for Windows (Git-2.19.0 at the time of writing). Run the installer. I just leave all the settings on the defaults but you are free to modify them if you like.

Note

If you choose to use the default MinTTY over the Windows default console window, be aware you need to preface all python commands with ‘winpty’ for example ‘winpty python manage.py migrate’

Installing Python 3

Go to http://www.python.org/download/ and download the latest Python 3.6.X (3.6.6 at the time of writing). Run the installer and on the first page, make sure you click the Add Python 3.6 to PATH option before choosing “Customize Installation”.

On the second page of the installer, leave the defaults and click “Next”.

On the third page, make sure you have “Install for all users” selected (this is important!) before clicking “Install”.

Warning

Python 3.7 is a recently released version of Python which is not compatible with Django 1.11 and hence QATrack+

Checkout the latest release of QATrack+ source code from BitBucket

Open a Windows PowerShell terminal and then create a directory for QATrack+ and check out the source code, use the following commands:

mkdir C:\deploy
cd C:\deploy
git clone https://bitbucket.org/tohccmedphys/qatrackplus.git

Setting up our Python environment

Ensure you have python3 installed correctly and on your PATH by running:

python --version
# should print e.g. Python 3.6.6

We’re now ready to install all the libraries QATrack+ depends on.

mkdir venvs
python -m pip install --upgrade pip
python -m venv venvs\qatrack3
.\venvs\qatrack3\Scripts\Activate.ps1
python -m pip install --upgrade pip
cd qatrackplus
pip install -r requirements\win.txt
python manage.py collectstatic

Warning

If you are going to be using Active Directory for authenticating your users, you need to install pyldap. There are binaries available on this page: https://www.lfd.uci.edu/~gohlke/pythonlibs/#pyldap. Download the binary relevant to your Python 3 installation (e.g. pyldap‑2.4.45‑cp36‑cp36m‑win_amd64.whl) and then pip install it:

pip install C:\path\to\pyldap‑2.4.45‑cp36‑cp36m‑win_amd64.whl

Checking everything is functional so far

Lets take a minute and check everything is now functioning as it should. Run the QATrack+ test suite like so:

py.test -m "not selenium"

This should take a few minutes to run and should exit with output that looks similar to the following:

Results (88.45s):
    440 passed
      2 skipped
     11 deselected

Creating a database with SQL Server

Open SQL Server Management Studio and enter ‘localhost’ for the server name and click Connect.

In the Object Explorer frame, right click the Databases folder and select “New Database…”.

Enter ‘qatrackplus’ as the database name and click OK.

Back in the Object Explorer frame, right click on the main Security folder and click New Login… Set the login name to ‘qatrack’, select SQL Server Authentication. Enter ‘qatrackpass’ (or whatever you like) for the password fields and uncheck Enforce Password Policy. Click OK.

Back in the Object Explorer frame, expand the qatrackdb database, right click on Security and select New->User.

Enter ‘qatrack’ as the User name and Login name and then in the Database Role Membership (or Owned Schemas) region select ‘db_datawriter’, ‘db_datareader’ and ‘db_owner’. Click OK.

Configuring QATrack+ to use your new database

Copy the example local_settings file:

cp deploy\win\local_settings.py qatrack\local_settings.py

and then edit it setting the DATABASES[‘default’][‘ENGINE’] key to sql_server.pyodbc:

DEBUG = False

DATABASES = {
    'default': {
        'ENGINE': 'sql_server.pyodbc',
        'NAME': 'qatrackplus',
        'USER': '',  # USER/PWD can usually be left blank if SQL server is running on the same server as QATrack+
        'PASSWORD': '',
        'HOST': '', # leave blank unless using remote server or SQLExpress (use 127.0.0.1\\SQLExpress or COMPUTERNAME\\SQLExpress)
        'PORT': '', # Set to empty string for default. Not used with sqlite3.
        'OPTIONS': {
        }
    }
}

ALLOWED_HOSTS = ['127.0.0.1', 'localhost']  # See local settings docs

We will load some configuration data into our new database from the command prompt:

python manage.py migrate
python manage.py createsuperuser
Get-ChildItem .\fixtures\defaults\*\*json | foreach {python manage.py loaddata $_.FullName}

Configuring CherryPy to Serve QATrack+

In order to have QATack+ start when you reboot your server, or restart after a crash, we will run QATrack+ with a CherryPy server installed as a Windows service (running on port 8080, see note below if you need to change the port).

cp deploy\win\QATrack3CherryPyService.py .
python QATrack3CherryPyService.py --startup=auto install
python QATrack3CherryPyService.py start

Your QATrack+ installation is now installed as a Windows Service running on port 8080 (see note below). You may also wish to configure the service to email you in the event of a crash (see the Recovery tab of the QATrackCherryPyService configuration dialogue).

Note

If you need to run QATrack+ on a different port, edit C:\deploy\qatrackplus\QATrack3CherryPyService.py and set the PORT variable to a different port (e.g. 8008)

Setting up IIS

We are going to use IIS for two purposes: first, it is going to serve all of our static media (css, js and images) and second it is going to act as a reverse proxy to forward the QATrack+ specific requests to CherryPy.

Before starting please make sure you have both URL Rewrite 2.0 and Application Request Routing 3.0 IIS modules installed.

Enabling Proxy in Application Request Routing

Application Request Routing needs to have the proxy setting enabled. To do this, click on the top level server in the left side panel, and then double click the Application Request Routing icon. In the Actions panel click the Server Proxy Settings and then check Enable proxy at the top. Leave all the other settings the same and click Apply and then Back to ARR Cache.

Enabling Static Content Serving in IIS

IIS is not always set up to serve static content. To enable this, open the Server Manager software, click Manage, then Add Roles and Features then Next, Next. In the Roles widget, select Web Server(IIS)->Common HTTP Features and make sure Static Content is selected.

Setting up the site and URL rewrite rules

Once you have Applicationn Request Routing installed and proxies enabled, in the left panel of IIS under Sites, select the default Web Site and click Stop on the right hand side.

Stop default website

Stop default website

Now right click on Sites and click Add Web Site

Add a new web site

Add a new web site

Enter QATrack Static for the Site Name and “C:\deploy\qatrackplus\qatrack" for the Physical Path then click OK and answer Yes to the warning.

To test that setup worked correctly open a browser on your server and enter the address http://localhost/static/qa/img/tux.png You should see a picture of the Linux penguin.

Next, select the top level server in the Connections pane and then double click URL Rewrite (you may need to restart IIS if you installed it and don’t see it here)

URL Rewrite

URL Rewrite

In the top right click Add Rule and select Blank Rule.

Give it a name of QATrack Static and enter ^(static|media)/.* for the Pattern field, and select None for the Action type. Make sure Stop processing of subsequent rules is checked.

Static Rule

Static URL Rewrite Rule

When finished click Apply, then Back To Rules and then add another blank rule. Give it a name of QATrack Reverse Proxy, enter ^(.*) for the Pattern and http://localhost:8080/{R:1} for the Rewrite URL. In the Server Variables section add a new Server Variable with the Name=HTTP_X_FORWARDED_HOST and the Value=yourservername.com (replace yourservername with whatever your domain is!). Finally, make sure both Append query string and Stop processing of subsequent rules are checked.

URL Rewrite Reverse Proxy

URL Rewrite Reverse Proxy

Your URL rewrites should look like the following (order is important!)

URL Rewrite rules

URL Rewrite rules

You should now be able to visit http://localhost/ in a browser on your server and see the QATrack+ login page. Congratulations, you now have a functional QATrack+ setup on your Windows Server!

Note

There are many different ways to configure IIS. The method I’ve used above is simple and works well when QATrack+ is the only web service running on a server.

What Next

Wrap Up

This guide shows only one of many possible method of deploying QATrack+ on Windows. It is very similar to what is used at The Ottawa Hospital Cancer Centre and it has proven to be a very solid setup. If you’re stuck with a Windows stack it will likely work for you too. Please post on the QATrack+ Google Group if you get stuck!

Upgrading from version 0.2.8

In order to upgrade from version 0.2.8 you must first uprade to version 0.2.9. If you hit an error along the way, stop and figure out why the error is occuring before proceeding with the next step! If you want assistance with the process, please post to to the Mailing List.

Open A Terminal & Activate your virtual environment

We will use Powershell for this, but feel free to use Git Bash (or plain old CMD) if you prefer. Open a Powershell window and and activate your existing virtual environment:

cd C:\deploy\
.\venvs\qatrack\bin\Activate.ps1

# or if you are using git bash then you need to do

cd /c/deploy/
source ./venvs/qatrack/bin/activate

Backing up your database

It is extremely important you back up your database before attempting to upgrade. It is recommended you use SQLServer Management Studio to dump a backup file, but you can also generate a json dump of your database (possibly extremely slow!):

cd C:\deploy\qatrackplus\
python manage.py dumpdata --natural > backup-0.2.8-$(date -I).json

Checking out version 0.2.9

First we must check out the code for version 0.2.9:

git fetch origin
git checkout v0.2.9.1

Warning

If you get any errors using git (e.g. trying to check out v0.2.9.1) that you don’t know how to handle, please stop and get help!

Update your existing virtual environment

There were a number of changes in dependencies for version 0.2.9 so we need to update our virtual env:

pip install --upgrade pip
pip install -r requirements\base.txt

Migrate your database

The next step is to migrate the 0.2.8 database schema to 0.2.9:

python manage.py syncdb
python manage.py migrate

Assuming that proceeds without errors you can proceed to Upgrading from version 0.2.9 below. If you get an error in this step, you may need to adjust your local_settings.py file to include the OPTIONS key in your DATABASES setting:

DATABASES = {
    'default': {
        'ENGINE': 'sqlserver_ado',
        'NAME': 'YOURDBNAME',
        'USER': '',
        'PASSWORD': '',
        'HOST': '',      # leave blank unless using remote server or SQLExpress (use 127.0.0.1\\SQLExpress or COMPUTERNAME\\SQLExpress)
        'PORT': '',                      # Set to empty string for default. Not used with sqlite3.
        'OPTIONS': {
            'provider': 'sqlncli11', # might need to use 'sqlncli10',
            'use_legacy_date_fields': True,
        }
    }
}

Upgrading from version 0.2.9

The steps below will guide you through upgrading a version 0.2.9 installation to 0.3.0. If you hit an error along the way, stop and figure out why the error is occuring before proceeding with the next step!

Verifying your Python 3 version

Unlike QATrack+ v0.2.9 which runs on Python 2.7, QATrack+ 0.3.0 only runs on Python version 3.5 or 3.6 (and probably 3.4!). You will need to ensure you have one of those Python versions installed. Instructions for installing Python 3.6 are given above. After installing Python 3 open a new PowerShell window and verify Python3 is installed correctly:

python --version
# should result in e.g.
Python 3.6.6

Note

If your python version says 2.7.x then you need to edit your PATH environment variable. Remove Python 2 paths and/or insert Python 3 paths.

Backing up your database

It is extremely important you back up your database before attempting to upgrade. It is recommended you use SQLServer Management Studo to dump a backup file, but you can also generate a json dump of your database (possibly extremely slow!):

cd C:\deploy\
.\venvs\qatrack\bin\Activate.ps1
cd qatrackplus\
python manage.py dumpdata --natural > backup-0.2.9-$(date -I).json

Checking out version 0.3.0

First we must check out the code for version 0.3.0:

git fetch origin
git checkout v0.3.0.9

Create and activate your new virtual environment

If you currently have a virtualenv activated, deactivate it with the deactivate command:

deactivate

We need to create a new virtual environment with the Python 3 interpreter:

cd C:\deploy
python -m pip install --upgrade pip
python -m venv .\venvs\qatrack3
.\venvs\qatrack3\Scripts\Activate.ps1

We’re now ready to install all the libraries QATrack+ depends on.

cd C:\deploy\qatrackplus\
python -m pip install --upgrade pip
pip install -r requirements\win.txt
python manage.py collectstatic

Warning

If you are going to be using Active Directory for authenticating your users, you need to install pyldap. There are binaries available on this page: https://www.lfd.uci.edu/~gohlke/pythonlibs/#pyldap. Download the binary relevant to your Python 3 installation (e.g. pyldap‑2.4.45‑cp36‑cp36m‑win_amd64.whl) and then pip install it:

pip install C:\path\to\pyldap‑2.4.45‑cp36‑cp36m‑win_amd64.whl

Update your local_settings.py file

Now is a good time to review your local_settings.py file. There are a few new settings that you may want to configure. The settings are documented in the settings page. Most importantly you need to update your database driver to use sql_server.pyodbc. Open your local_settings.py file and set the DATABASES[‘default’][‘ENGINE’] key to sql_server.pyodbc. If you had any OPTIONS keys set, you should remove those:

DATABASES = {
    'default': {
        'ENGINE': 'sql_server.pyodbc',
        'NAME': 'yourdatabasename',
        'USER': '',
        'PASSWORD': '',
        'HOST': '',
        'PORT': '',
        'OPTIONS': {
        }
    }
}

Migrate your database

The next step is to update the v0.2.9 schema to v0.3.0:

python manage.py migrate --fake-initial

and load some initial service log data:

Get-ChildItem .\fixtures\defaults\units\*json | foreach {python manage.py loaddata $_.FullName}
Get-ChildItem .\fixtures\defaults\service_log\*json | foreach {python manage.py loaddata $_.FullName}

Check the migration log

During the migration above you may have noticed some warnings like:

Note: if any of the following tests process binary files (e.g. images, dicom files etc) rather than plain text, you must edit the calculation and replace ‘FILE’ with ‘BIN_FILE’. Tests:

Test name 1 (test-1)
Test name 2 (test-2)

This data is also available in the logsmigrate.log file. Because the way Python handles text encodings / files has changed in Python 3, you will need to update any upload test that handles binary data by changing the FILE reference in the calculation procedure to BIN_FILE. For example change:

data = FILE.read()
# do something with data

to:

data = BIN_FILE.read()
# do something with data

You may have also seen warnings like:

The test named ‘yourtestname’ with ID=1234 needs to be updated to be
compatible with Python 3.

While most Test calculation procedures will be compatible with both Python 2 and Python 3, there have been some syntactical changes in the language which may require you to update a calculation procedure to be Python 3 compatible.

Update your CherryPy Service

First, stop your existing QATrack CherryPy Service using the Services Windows application. Then back in your PowerShell window you can install our new Python 3 CherryPy Windows Service:

cp deploy\win\QATrack3CherryPyService.py .
python QATrack3CherryPyService.py --startup=auto install
python QATrack3CherryPyService.py start

Your QATrack+ v0.3.0 installation is now running as a Windows Service on port 8080 (see note below). You may also wish to configure the service to email you in the event of a crash (see the Recovery tab of the QATrackCherryPyService configuration dialogue).

Note

If you need to run QATrack+ on a different port, edit C:\deploy\qatrackplus\QATrack3CherryPyService.py and set the PORT variable to a different port (e.g. 8008)

Once you have verified everything is working correctly, you can either disable the automatic startup of your original QATrackCherryPyService, or delete the service entirely.

IIS Changes

If you plan on using the QATrack+ API, you will want to modify your reverse proxy url rewrite and add a new Server Variable. Open up IIS and navigate to your reverse proxy rewrite rule and then in the Server Variables section add a new Server Variable with the Name=HTTP_X_FORWARDED_HOST and the Value=yourservername.com (replace yourservername with whatever your domain is!).

URL Rewrite Reverse Proxy

URL Rewrite Reverse Proxy

Last Word

There are a lot of steps getting everything set up so don’t be discouraged if everything doesn’t go completely smoothly! If you run into trouble, please get in touch on the mailing list.