AgentSkillsCN

performance-profiling

pprof 的使用与性能分析。在对应用程序进行性能剖析时使用。

SKILL.md
--- frontmatter
name: performance-profiling
description: pprof usage and performance analysis. Use when profiling applications.

Performance Profiling Skill

Using pprof for performance analysis.

When to Use

Use when identifying performance bottlenecks.

CPU Profiling

bash
# During tests
go test -cpuprofile=cpu.out -bench=.

# Analyze
go tool pprof cpu.out

# Web interface
go tool pprof -http=:8080 cpu.out

Memory Profiling

bash
# During tests
go test -memprofile=mem.out -bench=.

# Analyze
go tool pprof mem.out

HTTP Endpoint

go
import _ "net/http/pprof"

func main() {
    go func() {
        log.Println(http.ListenAndServe("localhost:6060", nil))
    }()
    // your code
}

Access profiles at:

  • •http://localhost:6060/debug/pprof/
  • •http://localhost:6060/debug/pprof/heap
  • •http://localhost:6060/debug/pprof/goroutine

pprof Commands

code
(pprof) top      # Top functions
(pprof) list     # Source code view
(pprof) web      # Graph view
(pprof) pdf      # PDF output

Best Practices

  • •Profile in production-like conditions
  • •Profile before optimizing
  • •Focus on hot paths
  • •Check both CPU and memory
  • •Use flamegraphs for visualization