example-poll-backend

This repo stores a minimal backend that works with videogular-questions to store and return poll information.

It is to be used as a guide to the polls API. The backend database and any processing that occurs after receipt, or prior to sending, is an example and can be switched out for whatever other system is required.

API

The server exposes two endpoints, a POST /vote and a GET /results/annotationId/questionId. Additionally some work is done to fix CORS problems that could occur.

The POST endpoint expects a schema similar to the vg-questions webworker:

{
  questionResult: questionId,
  annotation: annotationid
  result: ...
}

The GET endpoint returns an object of result<->vote-count pairs:

{
  optionOne: 5,
  cheese: 2
  last: 3
}

Example Setup

Starting the poll server can be done with python poll_server.py

Before use you should browse to /setup to allow the server to setup the database

Configuring an Apache webserver to run this application

First deploy this repository to the webserver.

Then the apache httpd.conf file needs to be configured by adding the following entry somewhere in the file:

<VirtualHost <hostname>:80>
	ServerName <hostname>
	WSGIDaemonProcess poll_server user=<user> group=<group> threads=5
	WSGIScriptAlias / /<location>/poll_server.wsgi
	ErrorLog logs/poll_server-error_log
	CustomLog logs/poll_server-access_log common

	<Directory <location>>
		WSGIProcessGroup poll_server
            WSGIApplicationGroup %{GLOBAL}
            Order deny,allow
            Allow from all
	</Directory>
</VirtualHost>

Parameters needing changes:

• <hostname> is the hostname of the server. e.g. hostname.domain.com
• <location> is the location of the sourcecode on the server. e.g. /var/www/poll_server/
• <user> is the user you want the script to run under, by default apache
• <group> is the group you want the script to run under, by default apache
• The ErrorLog and CustomLog parameters can be changed to any location

Unit Tests

We have written unit tests to ensure the poll server is correctly working.

These can be run with nosetests

Stress Testing

To run the stress testing script you need to install Locust.

Once this has been done you need to start the poll server and then start locust with the command

locust -f stress_testing.py --host http://127.0.0.1:5000

Browsing to localhost:8089 will show you the locust settings to begin the stress test.

This was tested using 50 users and found to have an average response time of 70 milliseconds.