Ensure the init script for each container is executable by running the following command:

$ sudo chmod -R 777 internet-scale-app_00X

Ensure that a mysql container with a database named cs4260 and a user 'www'@'%' who is granted all privileges to cs4260 and test_cs4260 (the test database Django test Client will create later). Otherwise, docker-compose up will not bring up any container

Add mysql container to docker networks backend by running the following command:

$ sudo docker network connect internet-scale-app_backend mysql

Depending on your docker version, network may be named to internetscaleapp_backend

Run sudo docker-compose up in project root folder to bring up docker containers

Note: it is possible that spark-worker will try to connect to spark-master multiple times before it forms a cluster with it. This is OK since spark (and most other distributed computing frameworks) will always attempt connecting until it forms a cluster.

Head to localhost:8006/homepage to access the project:

• After docker-compose is fully started, run the advanced fixture with the following command:

$ sudo docker exec -it presentation /bin/bash
$ python fixture.py

If successful, running the fixture will produce the following output:

The fixture will register a user with username test_user and password ABC123456789 and will then create 3 pets. After that, it will view each pet 5 times (cache-free)

• Run spark script to start recommendation service by running the following command (in a separate terminal):

$ sudo chmod 777 data/update_recommendation.sh
$ ./data/update_recommendation.sh

Recommendations will be updated every 60 seconds

• To create a new pet, click [Register] to register as a new user first
• After registration, user will be redirected to login page. Click [Log in] after filling in user credentials. A user who is already logged in will be redirected to homepage
• After logging in, click [Create a new pet!] to create a new pet
• After a new pet is created, user will be redirected to homepage
• Click [Check it out!] on each pet created to view its detailed information
• Since pet details are cached in redis container (cache will be invalidated every 20 minutes / when a user logs out). When a pet is cached, accessing it detail page will not be logged (as a result, its recommendation feed will not be updated either). Therefore, to test out the recommendation service, it is recommended to create multiple users / log out and re-log in again between viewing each pet. Only co-viewed pet with 3 or more views will be considered as a recommendation. Recommendations are updated every 60 seconds by running update_recommendation.sh. An example of recommendation feed produced from pre-loaded fixture data looks like as follows:

• To test out the updated search bar, simply type in it: suggestions from search history will appear (if the user has one) as follows:

Note: the suggestion panel will not show if input does not match any item from search history or when it is the first time a user searches (i.e. when there is no search history)

• If the suggestion panel is blocking any content on the page (for example: the login and register page), simply click outside of the suggestion panel to clear it