Load Balancer 🚀

This project implements an advanced HTTP load balancer in Go written according to a specific use case (randomized round-robin). The load balancer forwards incoming HTTP requests to a list of primary RPC (Remote Procedure Call) endpoints. If all primary endpoints fail, it falls back to a secondary list of RPC endpoints. Additionally, it includes a caching mechanism to temporarily avoid endpoints that return a "429 Too Many Requests" status and a notification system to alert when all endpoints are down. 📡

Architecture 🏗️

Logic Overview 🧠

1. Initialization:

   • The .env file is loaded to fetch the primary and fallback RPC URLs.
   • The URLs are split into slices and stored in global variables.
   • The application fails to start if no primary or fallback RPCs are provided. ❌

2. Main Function:

   • Initializes a cache to store URLs that return a "429 Too Many Requests" status temporarily using an in-memory database. 🗃️
   • Starts an HTTP server that listens for incoming requests. 🌐

3. Load Balancer:

   • Handles incoming HTTP requests and attempts to forward them to the primary RPCs. 📬
   • If all primary RPCs fail, it tries the fallback RPCs. 🔄
   • If both primary and fallback RPCs fail, it sends a notification to Slack and returns an error response. 🚨

4. Request Forwarding:

   • Shuffles the list of URLs to distribute the load. 🔀
   • Forwards the request to each URL until a successful response is received. ✅
   • If a URL returns a "429 Too Many Requests" status, it is cached in a Two-Queue (2Q) in-memory database to avoid retrying it for a specified TTL (time-to-live). ⏳

5. Caching Mechanism:

   • Uses a Two-Queue (2Q) in-memory caching algorithm to store URLs that return a "429 Too Many Requests" status. 🗄️
   • The cache temporarily stores URLs that return a "429 Too Many Requests" status, preventing repeated attempts to the same endpoint that has reached its rate limit. 🚫
   • ERROR_TIME_TO_LIVE_MINUTES is the time-to-live (TTL) for each URL in the cache. ⏲️

6. Slack Notification:

   • Sends a notification to Slack when all RPC endpoints fail to SLACK_WEBHOOK_URL. 📩

Docker 🐳

The Dockerfile is used to build a lightweight Docker image for the load balancer. It follows a multi-stage build process to minimize the final image size:

1. Builder Stage:

   • Uses the official Go image as the base image.
   • Copies the source code into the image.
   • Builds the Go binary for Linux using go build.

2. Runner Stage:

   • Uses the scratch image as the base, which is an empty image.
   • Copies the compiled binary from the builder stage.
   • Exposes the specified port.
   • Sets the command to run the binary when the container starts.

Example .env file:

ENV=production
RPCs=http://rpc1.com,http://rpc2.com,http://rpc3.com
FALLBACK_RPCs=http://fallback1.com,http://fallback2.com
PORT=8080
ERROR_TIME_TO_LIVE_MINUTES=5
SLACK_WEBHOOK_URL=https://hooks.slack.com/services/XXXXXXXXX/YYYYYYYYY/

This multi-stage build process ensures that the final image only contains the compiled binary and its runtime dependencies, resulting in a highly optimized and small image size.

Docker Compose Setup 🐙

The docker-compose.yml file is used to define and run the load balancer service using Docker Compose. It includes the following configuration:

• Service Definition:

  • Specifies the service name as load-balancer.
  • Sets the build context and Dockerfile location.
  • Assigns the image name as loadbalancer.

• Environment Variables:

  • Sets the ENV environment variable to production.
  • Loads additional environment variables from the .env file.

• Port Mapping:

  • Maps the container port to the host port specified in the PORT environment variable.

• Restart Policy:

  • Configures the service to restart unless stopped manually.

To start the load balancer using Docker Compose, run the following command:

docker-compose build
docker-compose up [-d]

Example

To test the load balancer, you can use curl http://localhost:8080 to send an HTTP request to the load balancer. The load balancer will forward the request to the primary and fallback RPC endpoints based on the configuration. 🧪