Go Performance Optimization Guide: Mastering Go pprof

Go Performance Optimization Guide: Master Go pprof The Go language provides the pprof performance analysis tool, which can be optimized through the following steps: Install the Go pprof tool: go install golang.org/x/perf/cmd/pprof@latest to generate performance analysis File: pprof -seconds=5 http-server.exe analysis Performance profiling file: pprof http-server.exe, use the top, flamegraph, callgraph and web commands in the command line tool to analyze memory, CPU and time. Practical case: Fix memory leaks by optimizing code to release *http.Handler instances, significantly improving program efficiency.

Go Performance Optimization Guide: Mastering Go pprof

Introduction

Go language provides a powerful performance analysis tool pprof, which can help Developers identify and optimize program performance bottlenecks. This article will introduce how to use pprof for performance analysis, and demonstrate how to optimize Go code through a practical case.

Installation and use of pprof

1. Install Go pprof tool:

go install golang.org/x/perf/cmd/pprof@latest

2. Generate performance profiling file:

pprof -seconds=5 http-server.exe

Analyze performance profiling files

The files generated by pprof can be viewed and analyzed using interactive command line tools:

pprof http-server.exe

Common commands

• top: Display the functions that consume the most time, CPU, and memory.
• flamegraph: Generate a flame graph to visualize the time distribution of function calls.
• callgraph: Generate a call graph to show the calling relationship between functions.
• web: Launch an interactive web interface that provides all analysis functions.

Practical Case: Memory Optimization

The following is a simplified Go code example, which will have a memory leak problem:

package main

import (
    "fmt"
    "runtime"

    "github.com/gorilla/mux"
)

func main() {
    router := mux.NewRouter()
    router.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {})

    // 在处理请求中创建大量的空结构体
    for i := 0; i < 10000; i++ {
        router.HandleFunc(fmt.Sprintf("/%d", i), func(w http.ResponseWriter, r *http.Request) {})
    }
}

Use pprof to analyze memory leaks

1. Run the program and generate a memory profiling file:

pprof -allocspace=1024 http-server.exe

2. Use interactive command line tools to analyze the file:

pprof http-server.exe

3. Enter the following command to view the most allocated memory:

top -sort=inuse

The output will show the most allocated object types, such as:

Flat profile: alloc space = 1.05 GiB
   Node count   Node alloc bytes
               1       1.05 GiB       <nil>

This indicates that the program generates a large number of *http. Handler instances, which are never released after handling the request.

Optimize the code

To fix the memory leak, you can modify the code and manually release it when no longer needed *http.Handler Example:

for i := 0; i < 10000; i++ {
    path := fmt.Sprintf("/%d", i)
    router.Handle(path, http.HandlerFunc(myHandler)).Methods(http.MethodGet)
    runtime.SetFinalizer(&router.Handlers[path], func(h http.Handler) {
        h.ServeHTTP(nil, nil)
        router.Remove(path)
    })
}

Optimization Results

Rerunning the program using the optimized code and generating a memory profiling file, it can be observed that the memory leak has been resolved:

Flat profile: alloc space = 20 MiB
   Node count   Node alloc bytes
               1        20 MiB       <nil>

Conclusion

pprof is a powerful tool , which helps developers identify and optimize the performance of Go code. By understanding how to generate and analyze pprof profiling files, developers can quickly identify and resolve performance issues, significantly improving program efficiency.

The above is the detailed content of Go Performance Optimization Guide: Mastering Go pprof. For more information, please follow other related articles on the PHP Chinese website!