Microservices Dotnetcore Docker Sf K8s Versions Save

• Adjust manifests to deploy in a cloud cluster as a cloud-native solution
• Deployed, ran and tested on Google Kubernetes Engine (GKE)

• Add health checks to all microservices
• Add health checks UI add-on to frontend
• Upgrade EF Core to 3.0.0
• Organize better the Startup files
• Set up k8s probes
• Move environment variables to a common ConfigMap

• Add Duber.Infrastructure.EventBus.Idempotency project to handle idempotency at integration events level.
• Make TripFinishedIntegrationEvent idempotent in order to avoid it can be paid more than once due to concurrency, retries, etc.

• Create an independent service to manage the notifications in order to decouple it from the frontend and to allow a better scaling out for both, frontend and notifications service.
• Add Redis to the cluster in order for SignalR to work properly into the cluster.
• Disable Sticky Sessions in frontend's Ingress to allow a better load balancing since the notifications don't depend on the frontend anymore. The notification's Ingress is the one that has Sticky Sessions to manage the SignalR connections.

Kubernetes support:

• Enable the solution to being deployed on a local cluster
• Use an Nginx Ingress Controller to expose frontend (Trip and Invoice services optional if you want to expose the API's)
• Set up Nginx LB to use Sticky sessions in order for SignalR to work properly.

Frontend client dependencies:

• Use Libman to manage the client dependencies.
• Delete static SignalR client dependencies, using Libman now.

General enhancement:

• Refactor SignalR messaging in order to send messages only tho the connected client rather than all clients.
• Refactor RabbitMQ client in order to use named Queues and full support to async handlers.