The server component to the TourTrak system built using Django on Ubuntu 14.04.

###Dependencies Note: These will be installed by a script in the Installation instructions below

• Python 2.7
• Pip for Python2
• python-django version 1.6.11
• apache server
• mod_wsgi
• postgresQL
• PostGIS
• postgresql-server-9.3
• psycopg2
• binutils
• gdal-bin
• libproj-dev
• memcached
• rabbitmq-server

(This is where you begin the server setup instructions)

1. Update aptitude

   sudo apt-get update

2. Install postgreSQL

   sudo apt-get install postgresql

3. After, install postGIS, which is a postgreSQL extention for handling spatial data

   sudo apt-get install postgresql-9.3-postgis-2.1

4. Setup the postgreSQL database:

• switch to the default postgresql user by running

  sudo -iu postgres

• create a user w/ read and write permissions:

  createuser --pwprompt dev

  • Note: Change "dev" to what you want the database username to be
  • Note: The password cannot be blank

• Create the DCS database used to collect rider information:

  createdb DCS

• Setup the postGIS functions:

  psql DCS
  CREATE EXTENSION postgis;
  CREATE EXTENSION postgis_topology;
  \q

• Sign out of postgresuser

  exit

##Install The Application

1. Install git

   sudo apt-get install git

2. Create a 'devcycle' directory in /usr/local/

   sudo mkdir /usr/local/devcycle

3. Clone this repository into /usr/local/devcycle

   sudo git clone https://github.com/tourtrak/devcycle-server.git /usr/local/devcycle

4. IMPORTANT: The directory must be named 'devcycle' and not 'devcycle-server'. Django does not support hyphen names for it's applications.

5. Inside the root directory of your application, install all project dependencies by running our setup script

   cd /usr/local/devcycle

   sudo bash setup.sh (This will take a significant amount of time on mod_wsgi-httpd, 10-20 mins)

6. Create a virtual host & WSGI file for the Apache server to display the Django application by creating a new .conf file from the template.

sudo cp /usr/local/devcycle/001-devcycle.conf.template /etc/apache2/sites-available/001-devcycle.conf
sudo ln -s /etc/apache2/sites-available/001-devcycle.conf /etc/apache2/sites-enabled/001-devcycle.conf

7. Create and Change the application settings:

sudo cp /usr/local/devcycle/dataCollection/settings.py.template /usr/local/devcycle/dataCollection/settings.py
sudo vim /usr/local/devcycle/dataCollection/settings.py

• Under DATABASES, modify USER and PASSWORD fields to reflect the database user you created in "Install The Database".
• Find the Lines with SECRET_KEY, KEY, and SALT. Generate values for these options.
• Change "TIME_ZONE" to match the timezone of the server in “tz database” format (e.g. ‘America/New_York’)
• Change "DEBUG" to True if you are not in production
• Change "DEFAULT_MAP_LAT" and "DEFAULT_MAP_LON" to match the latitude and longitude, respectively, that maps in the dashboard should default to.
• Change "MAP_TILE_SERVER" to match the hostname of the server to retrieve map tiles from (currently configured to use MapQuest's free OSM tile server)

8. Restart the apache server to put all changes into effect.

sudo /etc/init.d/apache2 reload

###Setup The Database Migrate From South: If you are using an older version of the devcycle database, you will need to migrate away from south as it no longer exists in Django 1.9+. You can find instructions to do so here.

New Installation: At this point the database schema for the Server does not exist yet. We will use Django to add it. Django will look at the current models to set-up the schema that the Server requires.

Note all commands below need to be ran within the root directory of the Django Project (/usr/local/devcycle)

These commands will create and then apply the migrations:

  sudo python manage.py makemigrations
  sudo python manage.py migrate --fake-initial

• Note: this may prompt you to create a django auth user. Follow that process as well if it does.

Restart server again.

sudo service apache2 reload

• Import all static files.

sudo python manage.py collectstatic

Restart server again:

sudo service apache2 reload

##Devcycle Load Test Framework

The following refers to the Load Test Framework for the Server. The framework is written in ruby and is command line based. A configuration file is used to configure the load test(s). Please see config_sample.txt for formatting and usage.

The framework is primairly based on 2 ruby gems, typhoeus and faraday. Typhoeus is used for making parallel http requests to simulate traffic. Faraday is a http wrapper to make it easier to create http requests.

This can be used to simulate up to 200 concurrent http requests at a single time based on Typhoeus. To increase that number you must run the framework simultaneously on multiple machines.

Framework works for ruby version 1.9.3 have not tested for other versions of ruby.

1. gem install typhoeus

2. gem install faraday

/Main.rb - the main file to run the system

framework/load_test.rb - contains most of the logic and processing

jsons/cpu_load.json - location_update request json to send to the server

configs/config_sample.txt - Configuration Sample w/ example format

framework/response_handler.rb - Response Handler for the Http Requests

framework/ - contains the logic of the framework

configs/ - contains all the config files

jsons/ - contains all json files. The json files constitute the data you are sending. You specify which json file in the config file

If you want to write your own configurations then include them in the configs/ directory. If you want to write your own json then include them in the jsons/ directory.

It is important to note that you do not need specify the full relative paths of the config files when running the framework on the command line. The framework will search for all config files in the configs/ directory. When specifying the json files in the config files, you do not need to specify the full relative paths for the json files, just the name of the file. The framework looks automatically in the jsons/ directory for all json files.

Sample command

ruby Main.rb config_sample.txt

or

./Main.rb config_sample.txt

Configs are formatted as follows:

URL (the connection URL)
#Requests (max 200)
JSON (the name of the JSON object to pull data from)
- (terminates the test definition)

Example config:

http://centri-pedal2.se.rit.edu/join_group/groupCode/riderId/
200
join_group.json
-

Main_GET.rb allows for groupCode and riderId to be specified as hooks in the URL for data to be plugged into dynamically.

A corresponding JSON object would then look like this:

{
	"groupCode" : "BIG1",
	"riderId" : 15
}

You can also use random test data by specifying the groupCode or riderId to be "random". Here is an example JSON for a randomized test:

{
	"groupCode" : "random",
	"groupCodeBase" : "BIG",
	"groupCodeMin" : 1,
	"groupCodeMax" : 50,
	"riderId" : "random",
	"riderIdMin" : 2,
	"riderIdMax" : 100
}

Random group codes are generated by combing a groupCodeBase ("BIG" in the above example) with a random number. The above JSON would result in a random selection between BIG1, BIG2, BIG3, ..., BIG50.

Random rider IDs are selected from the specified range of values.

To look at the Server performance in real time the linux command top is used. When running the load tests it is important to record the performance of the server. When you run the load test make sure you run top on the server simultaneously in order to record the performance of the server at the time of the tests. By using this command and piping the results this is achievable to analyze the data. Then after recording, grep the text in order to get the parameters you want then simply graph the numbers.

The command below pipes the top command every 1 second into a text file called example. top -b -d 1 > example.txt

The command below then greps for only the CPU performance numbers cat example.txt | grep Mem | cut -c 8-53 | nl -i 1

The command below then greps for only the Mem performance numbers cat example.txt | grep Cpu | cut -c 35-39 | nl -i 1

You may encounter an error about a missing libcurl.dll when you attempt to run the load testing framework. Follow these steps to resolve it:

Go to this URL: http://rubyinstaller.org/downloads/

Download DevKit-tdm-32-4.5.2-20111229-1559-sfx.exe

Follow these instructions to set the devkit up on your machine: https://github.com/oneclick/rubyinstaller/wiki/Development-Kit

Go to this URL: http://www.paehl.com/open_source/?CURL_7.39.0&PHPSESSID=b155d4a69c7aa26b1fd67e7201d4fb3e

Select "Download libcurl.dll (all versions) only"

Extract the contents of the archive, copy libcurl.dll from the SSL folder, and paste it into your Ruby /bin folder (probably C:\Ruby193\bin if you're on Windows)

#####Connecting to the database after setup

1. Type sudo -iu postgres
2. Type psql DCS

#####Some helpful commands include:

1. \dt to list all tables
2. \q to quit
3. psql DCS < /path/to/script/script.sql for running scripts without having to connect

When you make changes to the model, you need to create migrations to reflect the changes in the database. You can do so by running this command:

sudo python manage.py schemamigration MODEL_NAME NAME_OF_MIGRATION --auto

• change "MODEL_NAME" to the model you changed ex: "rider"
• change "NAME_OF_MIGRATION" to what you want to call it.

This commands will then apply the newly created migrations to the database:

sudo python manage.py makemigrations

Now you need to reload the server

sudo service apache2 reload