running-mobile-gaming-examples.md

Running Beam Mobile Gaming Examples

Google Cloud provides free trials so these examples can be executed in Dataflow without cost.

• Links
• Prerequisites
• Running the Demos
  • User Score
  • Hourly Team Score
  • Leaderboard
  • Game Stats
  • Stateful Team Score
• Demo Profiling

Links

• Beam Mobile Gaming Examples are documented here: https://beam.apache.org/get-started/mobile-gaming-example/
• Java implementations are here: https://github.com/apache/beam/tree/master/examples
• Clojure/thurber ports of each example can be found in ../demo/game; this page documents how to deploy these demos.

Prerequisites

Before attempting to execute thurber demos in Dataflow, you must:

1. Create thurber-demo GCP project.
   • Alternatively you can use your default project or any other project you create; just replace references to thurber-demo project in this documentation with the project you choose to use.
2. Create gs://thurber-demo/ bucket in GCS (Google Cloud Storage) in the project you intend to execute the examples.
3. Enable Dataflow and Dataflow API in GCP.
   • (If you don't know how to do this, you'll get an error trying to deploy these demos which will give you a link to do this.)
4. Setup a GCP Service Account
   • You'll need to setup a service account to run these examples.
   • This involves ensuring a GOOGLE_APPLICATION_CREDENTIALS environment var is set before deploying any example.
5. Build demo uberjar

   • To run any of these examples in Dataflow you'll need to build a deployable uberjar containing the code:

       lein with-profile +demo,+dataflow uberjar
     

   • Note: the resultant JAR contains our non-AOT'd Clojure sources that will dynamically compile and execute in Dataflow!

User Score

• Follow Prerequisites above.

  • You may need to adjust --project and --region parameters in accordance with your GCP account:

      lein with-profile +demo,+dataflow run -m game.user-score/demo! \
            --appName="thurber-demo-user-score" \
            --jobName="thurber-demo-user-score-$(date +%s)" \
            --runner=DataflowRunner \
            --region=us-central1 --project=thurber-demo \
            --gcpTempLocation=gs://thurber-demo/gcp-temp \
            --filesToStage=$(ls target/*-standalone.jar)
    

Hourly Team Score

• Follow Prerequisites above

• Deploy the pipeline:

    lein with-profile +demo,+dataflow run -m game.hourly-team-score/demo! \
      --appName="thurber-demo-hourly-team-score" \
      --jobName="thurber-demo-hourly-team-score-$(date +%s)" \
      --runner=DataflowRunner \
      --region=us-central1 --project=thurber-demo \
      --gcpTempLocation=gs://thurber-demo/gcp-temp \
      --filesToStage=$(ls target/*-standalone.jar)
  

Leaderboard

• Follow Prerequisites above

• Create topic

  • gcloud pubsub topics create --project thurber-demo thurber-demo-game

• Create dataset

  • bq mk --project_id thurber-demo --dataset thurber_demo_game

• Deploy the pipeline:

    lein with-profile +demo,+dataflow run -m game.leader-board/demo! \
      --appName="thurber-demo-leaderboard" \
      --jobName="thurber-demo-leaderboard-$(date +%s)" \
      --runner=DataflowRunner \
      --region=us-central1 --project=thurber-demo \
      --gcpTempLocation=gs://thurber-demo/gcp-temp \
      --filesToStage=$(ls target/*-standalone.jar)
  

• In a separate terminal window, generate streaming data:

    lein with-profile +demo,+dataflow run -m game.injector \
        thurber-demo thurber-demo-game none        
  

• Afterwards, clean up

  • gcloud pubsub topics delete --project thurber-demo thurber-demo-game
  • bq rm --project_id thurber-demo --dataset thurber_demo_game

Game Stats

• Follow Prerequisites above

• Create topic

  • gcloud pubsub topics create --project thurber-demo thurber-demo-game

• Create dataset

  • bq mk --project_id thurber-demo --dataset thurber_demo_game

• Deploy:

    lein with-profile +demo,+dataflow run -m game.game-stats/demo! \
      --appName="thurber-demo-game-stats" \
      --jobName="thurber-demo-game-stats-$(date +%s)" \
      --runner=DataflowRunner \
      --region=us-central1 --project=thurber-demo \
      --gcpTempLocation=gs://thurber-demo/gcp-temp \
      --filesToStage=$(ls target/*-standalone.jar)
  

• In a separate terminal window, generate streaming data:

    lein with-profile +demo,+dataflow run -m game.injector \
        thurber-demo thurber-demo-game none        
  

• Afterwards, clean up

  • gcloud pubsub topics delete --project thurber-demo thurber-demo-game
  • bq rm --project_id thurber-demo --dataset thurber_demo_game

Stateful Team Score

• Follow Prerequisites above

• Create topic

  • gcloud pubsub topics create --project thurber-demo thurber-demo-game

• Create dataset

  • bq mk --project_id thurber-demo --dataset thurber_demo_game

• Deploy:

    lein with-profile +demo,+dataflow run -m game.stateful-team-score/demo! \
        --appName="thurber-demo-stateful-team-score" \
        --jobName="thurber-demo-stateful-team-score-$(date +%s)" \
        --runner=DataflowRunner \
        --region=us-central1 --project=thurber-demo \
        --gcpTempLocation=gs://thurber-demo/gcp-temp \
        --filesToStage=$(ls target/*-standalone.jar)
  

• In a separate terminal window, generate streaming data:

    lein with-profile +demo,+dataflow run -m game.injector \
        thurber-demo thurber-demo-game none        
  

• Afterwards, clean up

  • gcloud pubsub topics delete --project thurber-demo thurber-demo-game
  • bq rm --project_id thurber-demo --dataset thurber_demo_game

Profiling

Add the following arg to any deployment to profile:

--profilingAgentConfiguration='{"APICurated":true}'

GCP will make profiled data available visualized in StackDriver.