Chat App with gRPC

This simple chat app covers these technologies.

• gRPC
• gRPC-Web
• Envoy
• Kubernetes

I hope this project will help you to understand how they work.

What the Chat App Looks Like?

Architecture

Directory Arch

$ tree -L 1
chat_app_with_grpc
├── Makefile                  # Useful commands packaged
├── README.md                 # Project Documents
├── client                    # gRPC-Web Client App: React.js x TypeScript
├── docker-compose.local.yml  # For local deploy
├── docker-compose.yml        # Used with minikube deploy
├── docs                      # Attachment files of README.md
├── envoy                     # Used for docker-compose.local.yml
├── k8s                       # K8s manifests
├── proto                     # .proto files
├── protoc                    # To compile protocol buffers in docker
├── python_grpc_client        # Not necessary. You can use instead of grpcurl
├── scripts                   # Useful commands packaged
└── server                    # gRPC Chat Server: Golang

9 directories, 4 files

How to Start?

You can run the client app and the chat server on your local machine.

macOS is highly recommended.

• Required Commands
  • minikube
  • kubectl
  • make
  • awk (usually pre-installed on macOS and Linux)
  • docker-compose
  • yarn or npm
• Optional
  • python
    • To use python_grpc_client instread of grpcurl
    • Quick start | Python
    • pip install grpcio grpcio-tools
  • grpcurl
    • curl for gRPC

# Make sure you're in the root directory of this project!!

# Start K8s Server
# - "minikube start"
# - "kubectl apply"
$ make k8s.apply

# Open another terminal window
# because $ minikube tunnel runs foreground!!

# Get an Access to LoadBalancer Service
# This runs foreground.
$ minikube tunnel

# In another terminal, run the commands below.

# Start Client App
$ make c.start

# Go to http://localhost
# The following command is available on macOS.
# On the other OS, you just have to open a browser.
$ open http://localhost

Stop Servers

# Stop Client App
$ make c.stop

# In the "minikube tunnel" terminal, hit Ctrl-c.
# Password will be required again.

# Delete minikube after "kubectl delete"
$ make m.delete

Useful Commands

# Connect to LoadBalancer Serices
# Password is the same as the login password.
$ minikube tunnel

# Check the IP Address of the LB Services
# CLUSTER-IP of lb-service is what we want.
# If it's <pending>, you might not have run "$ minikube tunnel".
$ kubectl get svc 
# [stdout]
# NAME                  TYPE           CLUSTER-IP      EXTERNAL-IP    PORT(S)                        AGE
# chat-server-service   ClusterIP      10.106.74.122   <none>         10000/TCP,10001/TCP            25m
# lb-service            LoadBalancer   10.107.70.75    10.107.70.75   80:31793/TCP,10001:31858/TCP   25m

# Tail Logs of the Pods with the Label of "app: chat-server"
# and the Container Name of "chat-server"
$ kubectl logs -l app=chat-server -c chat-server --tail=200 -f