Here's how to set up HomeAssistant for yourself using kubernetes (read all steps before deciding what to do):

Prerequisites

1. docker or img
2. minikube (or equivalent)
3. kubectl (or equivalent)
4. java 17 if running java subprojects outside of containers
5. shell environment (currently only supports bash/linux)

Setup

If using GCR (Google Container Registry) or GAR (Google Artifact Registry)

1. Make a copy of env.gradle.example and remove the .example extension.
2. Configure properties in env.gradle.
3. If using GCR (Google Container Registry) or GAR (Google Artifact Registry), download JSON key file for a service account that has access to the registry, place it at the root of this repo. If working with local images, skip this step.
4. Run ./gradlew fullClusterSetup.
5. Run ./mtunnel.sh to be able connect to the minikube cluster.
6. Once every pod has a state of "Running", run ./gradlew initDb. This will create the schema for the main server.
7. After initDb is done, run ./gradlew deltaRun to initialize the schema and load seed data.

After following these instructions, HomeAssistant should be ready for use. The default password using the JWT signing key of 'devsigningkey' is automatically inserted into the database. You can run an UpdatableBCrypt simulation with another signing key and update the record to change the password in order to log in and use the app (recommended).

• Connect to the server at http://localhost:8080 (unauthenticated, should have a secure proxy in front of it).
• Connect to the frontend at http://localhost:4200 (despite being the developer port for an angular UI, this is running in production mode and mapped to this port because 8080 is already exposed).
• Connect to kafdrop at http://localhost:9000 (unauthenticated, should have a secure proxy in front of it).
• Connect to postgres at http://localhost:5432 (or by using the connection string postgres://<username>:<password>@localhost:5432/hams_data) (unauthenticated, should have a secure proxy in front of it).

Tips/Troubleshooting:

• To tear down the whole thing to start anew, run minikube stop && minikube delete.

• To delete one deployment (like the app-server0 or frontend0 deployments for testing changes in kubernetes), run kubectl get deploy to list deployments, then run kubectl delete deploy <deployment_name>. To reapply using changes in the images, run ./gradlew reloadCluster.

• The pods may start out of dependency order. They will be continually restarted by the kube control plane until the dependencies are resolved. Pods do not get restarted until every other pod has finished its 'ContainerCreating' phase. Most commonly, the app-server pod(s) will fail initially because the postgres pod takes longer to create, and the kafka-broker pod will fail initially as well because zookeeper takes longer.

• It takes about 12 minutes for every pod to have a 'Ready' state. This is after pulling new images for all three kafka pods and the postgres pods. This can be shortened by manually storing docker images locally and pushing them to the minikube cluster once per cluster lifecycle by running minikube image load <image_name>.

Here is the current planned architecture of the project, in a diagram that showcases how each part might communicate with the others:

Here is an up-to-date view of the standard node angular ui:

Login screen:

Dashboard: