Metric - HTTP Server Example

Code Source: https://github.com/gogf/examples/tree/main/observability/metric/http_server

Description

This example demonstrates how to collect and monitor HTTP server metrics in GoFrame using OpenTelemetry and Prometheus integration. It shows how to:

• Monitor HTTP server requests
• Track request latencies
• Collect error rates
• Export server-side metrics

Structure

- `go.mod`: The Go module file for dependency management
- `go.sum`: The Go module checksums file
- `main.go`: The main application demonstrating HTTP server metrics

Features

The example showcases the following endpoints and metrics:

1. Endpoints

   • /: Basic endpoint returning "ok"
   • /error: Endpoint that triggers an error
   • /sleep: Endpoint with 5-second delay
   • /metrics: Prometheus metrics endpoint

2. Request Metrics

   • Total requests count
   • Active requests
   • Request duration
   • Request size

3. Response Metrics

   • Response status codes
   • Response size
   • Error count
   • Response latency

4. Server Metrics

   • Goroutine count
   • Memory usage
   • GC statistics
   • Connection stats

Requirements

• Go 1.22 or higher
• Git
• GoFrame
• GoFrame OpenTelemetry Metric

Usage

1. Run the example:

   go run main.go

2. Test different endpoints:

   # Basic request
   curl http://localhost:8000/
   
   # Error request
   curl http://localhost:8000/error
   
   # Slow request
   curl http://localhost:8000/sleep

3. View metrics:

   # Using curl
   curl http://localhost:8000/metrics
   
   # Or open in browser
   http://localhost:8000/metrics

4. Example metrics output:

   # HELP goframe_http_server_requests_total Total number of HTTP requests made
   goframe_http_server_requests_total{method="GET",path="/",status="200"} 1
   
   # HELP goframe_http_server_request_duration_seconds Duration of HTTP requests
   goframe_http_server_request_duration_seconds_bucket{method="GET",path="/sleep",status="200",le="5.0"} 1
   
   # HELP goframe_http_server_panics_total Total number of HTTP requests that caused a panic
   goframe_http_server_panics_total{method="GET",path="/error"} 1

Implementation Details

The example demonstrates:

1. HTTP server setup with metrics
2. Automatic metric collection
3. Built-in metrics integration
4. Error handling and tracking
5. Latency monitoring

Notes

• Metrics are collected automatically
• No manual instrumentation needed
• Built-in metrics are enabled by default
• Supports all HTTP methods
• Tracks all response codes
• Default port is 8000
• Metrics endpoint is at /metrics
• Consider security when exposing metrics
• High cardinality paths are normalized