gnunn1/summit-game-ansible

Playbook for provisioning the Red Hat Summit 2016 Balloon Game Demo

summit-game-ansible's Issues

Create rollback pipeline

It would be nice to have a rollback pipeline, this should just be a copy of the existing pipeline without the build step. i.e. just flip the colors.

Shorten provisioning time

Right now provisioning the demo takes about 15 minutes since it provisions and waits for each component serially. While this is required for the score, achievement and gamebus servers the nodejs applications (mobile-app, mobile-app-admin, leaderboard, scoreboard) and jenkins could be done in parallel to reduce the time.

nodejs applications failing

All of the nodejs applications are failing to run properly so something has changed since I last tried this demo. Not sure if it is something on the node side or something to do with the images.

Move state to JDG for separate blue/green clusters

Right now the demo uses a single cluster that spans the blue/green environments. This means that vertx cluster wide maps are shared between the two which is good for preserving state. However it also means that vertx eventbus messages can be served from either cluster which goes against being able to test them independently.

Moving the state (game state, playerNames and configuration) into an externalized cache like JDG would allow for completely separate clusters while preserving state.

Make admin credential configurable

Right now the admin credential is hard-coded in the vertx-game-server, it would be nice to make this configurable and readable from an environment variable or secret. Note the jenkins pipeline also needs the token to send the reconnect message.

Break gamebus up into multiple services

The gamebus uses multiple ports for handling the different traffic (leaderboards, scores, etc) and these need to be exposed as separate services in order to be routed outside of the cluster for the various nodejs applications.