revproxy-go

A reverse proxy server written in Go. It matches client endpoints to one of multiple backend URLs, selected based on customizable load-balancing strategies.

Features

• Load Balancing Strategies:

  • Round-robin
  • Weighted random
  • Pure random

• Ban System: Temporarily disables poorly performing backends based on:

  • Response status codes
  • Response body keyword matching

• SOCKS5 Proxy Support: Each backend can be optionally attached to a SOCKS5 tunnel.

• HTTPS/TLS Support: Configure certificates in the config file.

Usage

Go v1.22 or higher is required.

1. Clone the repository:

   git clone https://github.com/abswn/revproxy-go.git
   cd revproxy-go

2. Run tests:

   go test ./... -v

3. Build the binary:

   go build -o revproxy-go .

4. Run the server:

   ./revproxy-go

Directory Structure

revproxy-go/
├── cmd/                   # Future command-line tools
├── configs/
│   ├── config.yaml        # Main server configuration
│   └── endpoints/         # Per-site endpoint configurations
├── internal/
│   ├── ban/
│   ├── cert/
│   ├── config/
│   ├── forward/
│   └── strategy/
├── main.go
└── README.md

Configuration

Global Configuration (configs/config.yaml)

port: 44562

# Path to certs or leave empty to use without encryption 
https_cert_path: ""
https_key_path: ""

log:
  level: "info"             # Options: debug, info, warn, error, off
  output: "logs/output.log" # "stdout" or a file path like "logs/output.log"
  format: "text"            # "text" or "json"

Endpoint Configuration (configs/endpoints/example.yaml)

Multiple YAML files can be used to separate the endpoints logically. All the configs with enabled flag true will be active.

enabled: true

endpoints:
  "/api":
    strategy: round-robin # random, weighted, round-robin
    urls:
      - url: "https://example.com/api1"
        # Optional: socks5, username, password
        socks5: "127.0.0.1:1080"
        username: "user1"
        password: "pass1"
      - url: "https://example.com/api2"
      - url: "https://example.com/api3"
    ban:
      - match : ["429", "try after some time"]
        duration: 30 # temporarily disables a backend for 30 secs
      - match : ["500"]
        duration: 3600

  "/path":
    strategy: weighted
    urls:
      - url: "https://example.com/api1"
        weight: 50 # Required for weighted strategy
      - url: "https://example.com/api2"
        weight: 30
      - url: "https://example.com/api3"
        weight: 20

# Applies to all endpoints and is overridden locally
global_ban:
  - match: ["429"]
    duration: 30 # in seconds
  - match: ["503", "out of capacity"]
    duration: 3600

Explanation

• enabled: Whether this config file is active

• endpoints: Map of path to backend strategy and URLs

• strategy: Can be either round-robin, weighted or pure random

• urls: List of backend definitions

  • url: Backend target URL
  • socks5: Optional SOCKS5 proxy address
  • weight: Used only with weighted strategy

• ban / global_ban: The global_ban rules apply to all endpoints in the config. The local ban rules add to it or override it. Multiple keywords can be written in the same line.

Start the server and send requests.

curl http://localhost:44562/api

This will use one of the three backend servers to fetch the result. The selection of the backend server is done in round-robin format. If no healthy backends are available and all have been temporarily disabled, the server responds with 503 Service Unavailable.

Logging

Configured via config.yaml:

• level: debug, info, warn, error, off
• output: stdout or path to log file
• format: text or json

HTTPS/TLS Support

• If cert/key paths are provided, they are used.
• Otherwise fallback to non-encrypted HTTP.

License

MIT License