rotisserie

LIVE NOW

rotisserie takes the concept of the red zone in American football and applies it to the popular online battle royale game PLAYERUNKNOWN'S BATTLEGROUNDS. The idea is to always be viewing the most popular PUBG twitch stream with the least amount of people alive in-game.

Included Component

• Kubernetes Clusters

Featured Technologies

• Container Orchestration
• Microservices
• Node.js

Prerequisite

The following pieces of software are required to run rotisserie locally:

• tesseract-ocr
• ffmpeg
• imagemagick
• livestreamer

You can install these dependencies with one of the two following commands, depending on your os:

• Ubuntu:

  $ sudo apt-get install tesseract-ocr ffmpeg imagemagick
  $ pip install livestreamer

• macOS:

  $ brew install tesseract ffmpeg imagemagick
  $ pip install livestreamer

Create a Kubernetes cluster with either Minikube for local testing, or with IBM Bluemix Container Service to deploy in cloud. The code here is regularly tested against Kubernetes Cluster from Bluemix Container Service using Travis.

Steps

1. Get an OAuth Token for livestreamer
2. Setting up Environment Variables
3. Build the images
4. Deploy locally
5. Deploy using Docker
6. Deploy using Kubernetes

1. Getting an OAuth Token for Twitch

1. On a machine with a browser installed, run the following:

  $ livestreamer --twitch-oauth-authenticate

2. A browser window will open, and prompt you to authorize livestreamer to use your twitch account. Click Authorize.

3. Your browser will refresh, and a page saying "SORRY, this page does not exist yet" will appear. Ignore this. In your address bar, there will be a callback URL with access_token=<TOKEN>. This is your OAuth token, copy it down and proceed to the next section.

2. Setting up Environment Variables

• Create an environment variable for your docker username.

  $ export docker_username="YOUR_DOCKER_USERNAME"

• Create an enviornment variable for your token you retrieved in the previous step.

  $ export token="YOUR_OAUTH_TOKEN"

• Create an environment variable for your clientID.

You must create an app in your Twitch account using the developer mode (https://dev.twitch.tv/) to retrieve the clientID

  $ export clientID="YOUR_CLIENT_ID"

• Create an environment variable for the ROTISSERIE_OCR_SERVICE_HOST and ROTISSERIE_OCR_SERVICE_PORT. This can be set to localhost and 3001 if running locally, or the IP address and port of a remote OCR host if running in containers. Mimic the environment variables exported by kubernetes.

  $ export ROTISSERIE_OCR_SERVICE_HOST="localhost"
  $ export ROTISSERIE_OCR_SERVICE_PORT="3001"

3. Build the Images

• Clone the repo.

 $ git clone https://github.com/IBM/rotisserie.git
 $ cd rotisserie

• Build and Push the Docker Image. You would need to push it if you want to deploy the application in Kubernetes.

$ docker build -t $docker_username/rotisserie-ocr -f deploy/images/ocr.Dockerfile .
$ docker build -t $docker_username/rotisserie-app -f deploy/images/app.Dockerfile .
$ docker push $docker_username/rotisserie-ocr
$ docker push $docker_username/rotisserie-app

4. Running It Locally

• Install with npm.

  $ npm install .

• Navigate to the rotisserie dir if you aren't there already, and start the app:

  $ node ocr.js 2>&1 >/dev/null &
  $ node app.js

Now you can open a browser and navigate to http://localhost:3000 to watch rotisserie.

5. Running in a Container

You can also run rotisserie in a docker container.

• Start up the containers:

  $ docker run -d -p 3001:3001 --name rotisserie-ocr $docker_username/rotisserie-ocr
  $ docker run -d -p 3000:3000 --name rotisserie-app -e ROTISSERIE_OCR_SERVICE_HOST=$ROTISSERIE_OCR_SERVICE_HOST -e ROTISSERIE_OCR_SERVICE_PORT=$ROTISSERIE_OCR_SERVICE_PORT -e token=$token -e clientID=$clientID $docker_username/rotisserie-app

Now you can open a browser and navigate to http://localhost:3000 to watch rotisserie.

6. Running in Kubernetes

note: Ensure your $OCR_HOST environment variable is set to the cluster_public_ip:3001.

1. Create a Kubernetes Secret for your OAuth token. You will need to encode the data you want in Base64 for the Kubernetes Secret.

$ echo -n "YOUR_OAUTH_TOKEN" | base64

2. Modify the rotisserie-secrets.yaml file to use your token

...
data:
  token: YOUR_OAUTH_TOKEN_IN_BASE64

3. Finally, create the Kubernetes Secret.

$ kubectl create -f rotisserie-secrets.yaml

4. Modify the rotisserie-app.yaml and rotisserie-ocr.yaml yaml files to use your image.

...
    containers:
    - name: rotisserie-app
      image: <docker_username>/rotisserie-app

5. Deploy the OCR service then the main application.

$ kubectl apply -f rotisserie-ocr.yaml
$ kubectl apply -f rotisserie-app.yaml

• To access your application. You would need the public IP address of your cluster. If you don't have a load balancer, you can use the Node Port.

# For clusters provisioned with Bluemix
$ bx cs workers YOUR_CLUSTER_NAME

# For Minikube
$ minikube ip

• Now you can go to http://IP_ADDRESS:30080

Production Detail

The production version of rotisserie has slightly different operational procedures. The production kubernetes manifest is located in the deploy directory. It is typically interacted with via the provided Makefile. Major differences between the production rotisserie and the one used in the developer journey are use of ingress controllers in kuberenetes and adding Letsencrypt.

There are a few commands we can use to work with the deployment.

To deploy without letsencrypt use make roll

make roll

To deploy with letsencrypt use make full-roll

make full-roll

To redeploy the deployments, without removing svc/ing/other, use make redeploy

make redeploy

To delete the entire deployment we can use make purge

make purge

Note: this depends on you deploying with a unique sha. See the 'make-rev' rule in the Makefile. In most cases git pull; make roll should work. In cases where a roll failed or the app failed for reasons not connected to the code, a dummy commit might need to be added before re-rolling. Please only roll from master.

Whitelisting and Blacklisting streamers

To whitelist simply set an environment variable called ROTISSERIE_WHITELIST to a string with space separated usernames. Same with blacklisting, but with the environment variable ROTISSERIE_BLACKLIST.

License

rotisserie is currently licensed under the Apache 2.0 License.