A simple golang web server with basic logging, tracing, health check, graceful shutdown and zero dependencies

main.go

package main

import (
	"context"
	"flag"
	"fmt"
	"log"
	"net/http"
	"os"
	"os/signal"
	"sync/atomic"
	"time"
)

type key int

const (
	requestIDKey key = 0
)

var (
	listenAddr string
	healthy    int32
)

func main() {
	flag.StringVar(&listenAddr, "listen-addr", ":5000", "server listen address")
	flag.Parse()

	logger := log.New(os.Stdout, "http: ", log.LstdFlags)
	logger.Println("Server is starting...")

	router := http.NewServeMux()
	router.Handle("/", index())
	router.Handle("/healthz", healthz())

	nextRequestID := func() string {
		return fmt.Sprintf("%d", time.Now().UnixNano())
	}

	server := &http.Server{
		Addr:         listenAddr,
		Handler:      tracing(nextRequestID)(logging(logger)(router)),
		ErrorLog:     logger,
		ReadTimeout:  5 * time.Second,
		WriteTimeout: 10 * time.Second,
		IdleTimeout:  15 * time.Second,
	}

	done := make(chan bool)
	quit := make(chan os.Signal, 1)
	signal.Notify(quit, os.Interrupt)

	go func() {
		<-quit
		logger.Println("Server is shutting down...")
		atomic.StoreInt32(&healthy, 0)

		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
		defer cancel()

		server.SetKeepAlivesEnabled(false)
		if err := server.Shutdown(ctx); err != nil {
			logger.Fatalf("Could not gracefully shutdown the server: %v\n", err)
		}
		close(done)
	}()

	logger.Println("Server is ready to handle requests at", listenAddr)
	atomic.StoreInt32(&healthy, 1)
	if err := server.ListenAndServe(); err != nil && err != http.ErrServerClosed {
		logger.Fatalf("Could not listen on %s: %v\n", listenAddr, err)
	}

	<-done
	logger.Println("Server stopped")
}

func index() http.Handler {
	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/" {
			http.Error(w, http.StatusText(http.StatusNotFound), http.StatusNotFound)
			return
		}
		w.Header().Set("Content-Type", "text/plain; charset=utf-8")
		w.Header().Set("X-Content-Type-Options", "nosniff")
		w.WriteHeader(http.StatusOK)
		fmt.Fprintln(w, "Hello, World!")
	})
}

func healthz() http.Handler {
	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if atomic.LoadInt32(&healthy) == 1 {
			w.WriteHeader(http.StatusNoContent)
			return
		}
		w.WriteHeader(http.StatusServiceUnavailable)
	})
}

func logging(logger *log.Logger) func(http.Handler) http.Handler {
	return func(next http.Handler) http.Handler {
		return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			defer func() {
				requestID, ok := r.Context().Value(requestIDKey).(string)
				if !ok {
					requestID = "unknown"
				}
				logger.Println(requestID, r.Method, r.URL.Path, r.RemoteAddr, r.UserAgent())
			}()
			next.ServeHTTP(w, r)
		})
	}
}

func tracing(nextRequestID func() string) func(http.Handler) http.Handler {
	return func(next http.Handler) http.Handler {
		return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			requestID := r.Header.Get("X-Request-Id")
			if requestID == "" {
				requestID = nextRequestID()
			}
			ctx := context.WithValue(r.Context(), requestIDKey, requestID)
			w.Header().Set("X-Request-Id", requestID)
			next.ServeHTTP(w, r.WithContext(ctx))
		})
	}
}

Here's a git-push to deploy version of this with HTTPS domain: https://hasura.io/hub/projects/shahidhk/simple-go-server

In the tracing middleware, shouldn't you get the Request ID from the Context, rather than from the request Header.
Please, correct me if I am wrong!

It turns out, maybe a simple server isn't quite so simple, compared to the typical "Hello World."

Anybody want to discuss the use of goroutines and other techniques in this code? We're usually told that goroutines aren't necessary because this is already done in the standard library, for example.

It seems clear that this relates to resiliency and so on, but what caused the authors (original and contributing) to arrive at these conclusions and include additional checks and handling code, at the cost of increased complexity? There would appear to be context--likely based on knowledge and direct experience--that is somewhat lost in the example.

A breakdown of the code and decisions behind its parts would likely be very educational for those among us who wouldn't know to include these things by default.

if i started this server from the commandline like go run server.go, how do you stop it?

It turns out, maybe a simple server isn't quite so simple, compared to the typical "Hello World."

Anybody want to discuss the use of goroutines and other techniques in this code? We're usually told that goroutines aren't necessary because this is already done in the standard library, for example.

It seems clear that this relates to resiliency and so on, but what caused the authors (original and contributing) to arrive at these conclusions and include additional checks and handling code, at the cost of increased complexity? There would appear to be context--likely based on knowledge and direct experience--that is somewhat lost in the example.

A breakdown of the code and decisions behind its parts would likely be very educational for those among us who wouldn't know to include these things by default.

The real world is always messy, especially compared to programming examples on websites. It always helps to have experience with the platforms you're running on, code doesn't run in a vacuum.

When you run a web server, you need a process to listen for connections, and processes for handling the responses (which is handled behind the scenes by the Go standard library here). If you want to listen for signals from the system, you would need another process. This process just sleeps until it gets a signal, then calls the shutdown method on the web server. The main process then stops listening but needs to wait for that routine with shutdown to end. You can think of it as how many things you need to do concurrently, at the same time, and in which order they need to be handled.

This is pretty much how processes are handled in "traditional" unix software, so if you have experience with those platforms and the code running on them, adding this kind of functionality comes naturally.

The question is if you really are going to notice any problems if you wouldn't include graceful shutdown on a typical web server, where connections tend to be short-lived and services are rarely brought down...

if i started this server from the commandline like go run server.go, how do you stop it?

Ctrl-C (sigint), or use kill to send the process a signal.

how to stop the server once it started?

Anyone know how I can also log GET and POST arguments? Like for example printing out "index.php?firstname=John&lastname=Smith". Would be much appreciated!

Anyone know how I can also log GET and POST arguments? Like for example printing out "index.php?firstname=John&lastname=Smith". Would be much appreciated!

If you are interested in the GET query parameters, check the other fields and methods of URL, such as RequestURI, RawQuery or even just calling String() on URL itself.

If you also want the POST variables, you need to check out Request.Form and Request.ParseForm(). Note it's usually a bad idea to log POST request variables as they might be rather large or contain sensitive information such as passwords.

I added a -d option to specify a directory to serve out, but im no good with forking and pull requests. I think I might have removed some useful functionality in the process :)) but if you like that flag, you can find the repo pinned to my github and hack away at it

This is cash-money awesome! Thank you for sharing this!

Any pointers on how to include the http.statuscode in logs, please?

Thanks for sharing

I'd like to use this code in an open source static site generator. It's MIT licensed. Can I use this code, or at least a version of it? And if so, what's your license for it? Thanks!