Syncthing源码解析

我相信很多朋友会认为启动就是双击一下Syncthing程序图标,随后就启动完毕了!如果这样认为,对,也不对!对,是因为的确是这样操作,启动了Syncthing;不对是因为在调试Syncthing启动过程时发现很奇怪!有哪些奇怪的地方呢?

一,Syncthing启动的奇怪问题。

1,启动时候总是运行"monitorMain(options)"函数,另外一个函数"syncthingMain(options)"从来没有运行过?

2,启动log前缀好几种,第三种log前缀不知道什么意思?也不知道包含这些前缀的log信息从哪里来的?

3,每次Syncthing启动完毕,通过任务管理器查看,都有两个正在运行的Syncthing,怎么同时有两个Syncthing进程?

下面是上述问题的各个截图:

问题1截图:

问题2截图:

问题3截图:

 二,分析和解决Syncthing启动的奇怪问题。

我用Gogland调试时候发现,Syncthing启动总是进入"monitorMain(options)"函数,另外一个函数"syncthingMain(options)"从来没有进入过,难道另外一个函数彻底没用?没用还保留干什么?先不管这些了,我先找那些log前缀在哪里吧?

(1),[monitor]前缀,最终在monitor.go文件中的"monitorMain(options)"函数中发现,位于39行处。具体代码:

l.SetPrefix("[monitor] ")

(2),[start]前缀,没能在"monitorMain(options)"函数中找到!

(3),[NTUJ4]前缀,也没能在"monitorMain(options)"函数中找到!

难道这两个前缀的log信息,不是"monitorMain(options)"函数输出的?如果不是那会是哪个函数呢?经过调试发现,在"monitorMain(options)"函数中输出的最后一条log是:

[monitor] 17:34:20 INFO: Starting syncthing

继续调试,最终也没有发现剩余前缀log在哪里产生?但是确实千真万确输出这些前缀的log了呀!!!  

先不管上面问题了,我们还是去看看"syncthingMain(options)"函数吧,它真的就是一个多余的函数,彻底没用?为了测试这个函数是否有用,我在这个函数最开始代码处写入:

l.Infoln("这是我的log信息,是否会出现呢?")

Syncthing默认log全部是英文,我写的log信息是中文,如果这个函数被调用,那么我一眼就能看出来我写的log信息!!立即运行Syncthing,奇迹出现了,居然log信息中有我写的测试log信息!!!

"syncthingMain(options)"函数并不是没有用啊?我写的中文测试log居然在运行时输出的log中了!!其它log信息都在我写的中文log之后,难道剩余log信息都是它输出的?在这个函数中查找代码:

l.SetPrefix

最终发现两处:

(1),

l.SetPrefix("[start] ")

这个前缀不就是上面我没能找到的一个吗?

(2),

l.SetPrefix(fmt.Sprintf("[%s] ", myID.String()[:5]))

难道这是剩余的那个log前缀?从这段代码理解,应该是取当前设备ID的前5个字符作为log前缀!  

为了验证我的判断,我把这两个log前缀都修改了

l.SetPrefix("[syncthingMain][start] ")
l.SetPrefix("[syncthingMain] ")

修改完毕后,运行一下Syncthing,log前缀确实全变成我修改的了,我的判断完全正确!!!

经过测试发现,包含"[start]"前缀的log,只有在Syncthing第一次运行时候出现,系统当中从来没有运行过Syncthing才出现,如果运行过,删除先前的配置文件后也会出现,因为那样再次运行Syncthing,等于从来没有在此电脑中运行过Syncthing!

到现在为止,我也不知道"syncthingMain(options)"函数是怎么运行起来的,但是可以肯定它一定是运行了,从它输出的log来判断,它应该是负责启动之后的Syncthing各项服务!!!偶然通过任务管理器发现,Syncthing运行后,居然有两个正着运行的Syncthing,怎么会有两个进程?一个使用内存小;一个使用内存稍大(和前面的进程使用内存比较)!先不管了,我先终止一下这两个进程吧,继续调试Syncthing!奇怪事情又发生了,如果终止内存稍大的Syncthing,立即会再次运行和它一样的进程,就是杀不死!!!如果我终止使用内存小的Syncthing进程,另外一个内存稍大的进程依然存在,而且Syncthing的各项功能均运行正常!!!难道这两个进程一个是监控进程;一个是服务进程?

(1),监控进程:使用内存小,负责监控服务进程运行状况。

(2),服务进程:使用内存稍大,为用户提供Syncthing各项服务。

我们还是认真来看看源代码吧!

首先看monitor.go文件中的"monitorMain(options)"函数,经过调试我最终确定,是这个函数启动和监控Syncthing的服务进程!!!

(1),启动Syncthing的服务进程:在"monitorMain(options)"函数中,有一个for的死循环,在这个for代码前面部分,有一段是启动Syncthing的服务进程:

		cmd := exec.Command(args[0], args[1:]...)

		stderr, err := cmd.StderrPipe()
		if err != nil {
			l.Fatalln("stderr:", err)
		}

		stdout, err := cmd.StdoutPipe()
		if err != nil {
			l.Fatalln("stdout:", err)
		}

		l.Infoln("Starting syncthing")
		err = cmd.Start()
		if err != nil {
			l.Fatalln(err)
		}

(2),监控Syncthing的服务进程:上面那个for死循环,如果毫无节制,那系统肯定受不了啊?其实它是隔一段时间才执行一次的!看这个for循环最后一行的代码!

time.Sleep(1 * time.Second)

这行代码是休眠指定的时间,目前是一秒钟!!难怪Syncthing的服务进程被杀死,马上就会有另外一个Syncthing的服务进程运行起来,这样的设计,我一辈子也杀不死这个Syncthing的服务进程啊!!!!通过调整这个休眠时间测试,最终确定,这个地方就是控制监控Syncthing的服务进程的间隔时间代码!!!!  

到此,我可以肯定了,Syncthing启动时候,会先后启动两个进程:

(1),监控进程:

  •  最早运行。
  • 调用"monitorMain(options)"函数。
  • 使用内存小。
  • 负责监控服务进程运行状况。

(2),服务进程:使用内存稍大,为用户提供Syncthing各项服务。

  • 被监控进程启动。
  • 调用"syncthingMain(options)"函数。
  • 使用内存稍大。
  • 为用户提供Syncthing的各项服务。  

 继续调试"syncthingMain(options)"函数,证实我的判断是正确的,到此Syncthing的启动就算搞明白了!!

以后再调试Syncthing,不需要按照现在逻辑运行了,注释掉main.go中main函数末尾代码:

	if innerProcess || options.noRestart {
		syncthingMain(options)
	} else {
		monitorMain(options)
	}

把上面代码替换成:

syncthingMain(options)

这样我们就可以去掉监控进程,直接调试Syncthing服务进程了!!

最后还有一个问题,为什么Syncthing的监控进程里显示Syncthing服务进程输出的log?这是用GO语言的channel来完成的!!!

三,我调试时候,加入的注释的代码。

注释方式:采用TODO代码方式,这样便于在Gogland的TODO里面立即看到我自己加入的注释或修改代码!

注释格式:

TODO:[sunylat][begin或end]注释或修改描述信息
[begin或end]为可选项,表示需要相互协作才能工作代码的开始和结束处。
这个TODO可以用"//"和"/**/"两种注释方式包含。

(1),monitor.go文件中的"monitorMain(options)"函数:

/*
TODO:[sunylat] 运行和监控提供真正服务syncthing的入口函数
 */
func monitorMain(runtimeOptions RuntimeOptions) {
	os.Setenv("STMONITORED", "yes")
	l.SetPrefix("[monitor] ")

	var dst io.Writer = os.Stdout

	logFile := runtimeOptions.logFile
	if logFile != "-" {
		var fileDst io.Writer = newAutoclosedFile(logFile, logFileAutoCloseDelay, logFileMaxOpenTime)

		if runtime.GOOS == "windows" {
			// Translate line breaks to Windows standard
			fileDst = osutil.ReplacingWriter{
				Writer: fileDst,
				From:   '
',
				To:     []byte{'
', '
'},
			}
		}

		// Log to both stdout and file.
		dst = io.MultiWriter(dst, fileDst)

		l.Infof(`Log output saved to file "%s"`, logFile)
	}

	args := os.Args
	var restarts [countRestarts]time.Time

	stopSign := make(chan os.Signal, 1)
	sigTerm := syscall.Signal(15)
	signal.Notify(stopSign, os.Interrupt, sigTerm)
	restartSign := make(chan os.Signal, 1)
	sigHup := syscall.Signal(1)
	signal.Notify(restartSign, sigHup)

	/*
	TODO:[sunylat] 这是个无限循环,负责监控提供服务syncthing运行状况,同时负责处理程序的退出等事件处理,
	监控间隔时间由这个for循环最后面的休眠时间决定
	 */
	for {
		if t := time.Since(restarts[0]); t < loopThreshold {
			l.Warnf("%d restarts in %v; not retrying further", countRestarts, t)
			os.Exit(exitError)
		}

		copy(restarts[0:], restarts[1:])
		restarts[len(restarts)-1] = time.Now()


		//TODO:[sunylat] [begin] 启动提供服务的syncthing ----------------------
		cmd := exec.Command(args[0], args[1:]...)

		stderr, err := cmd.StderrPipe()
		if err != nil {
			l.Fatalln("stderr:", err)
		}

		stdout, err := cmd.StdoutPipe()
		if err != nil {
			l.Fatalln("stdout:", err)
		}

		l.Infoln("Starting syncthing")


		//TODO:[sunylat] 这条语句真正执行启动提供服务的syncthing
		err = cmd.Start()
		if err != nil {
			l.Fatalln(err)
		}
		//TODO:[sunylat] [end] 启动提供服务的syncthing --------------------------

		// Let the next child process know that this is not the first time
		// it's starting up.
		os.Setenv("STRESTART", "yes")

		stdoutMut.Lock()
		stdoutFirstLines = make([]string, 0, 10)
		stdoutLastLines = make([]string, 0, 50)
		stdoutMut.Unlock()

		wg := sync.NewWaitGroup()

		wg.Add(1)
		go func() {
			copyStderr(stderr, dst)
			wg.Done()
		}()

		wg.Add(1)
		go func() {
			copyStdout(stdout, dst)
			wg.Done()
		}()

		exit := make(chan error)

		go func() {
			wg.Wait()
			exit <- cmd.Wait()
		}()

		select {
		case s := <-stopSign:
			l.Infof("Signal %d received; exiting", s)
			cmd.Process.Kill()
			<-exit
			return

		case s := <-restartSign:
			l.Infof("Signal %d received; restarting", s)
			cmd.Process.Signal(sigHup)
			err = <-exit

		case err = <-exit:
			if err == nil {
				// Successful exit indicates an intentional shutdown
				return
			} else if exiterr, ok := err.(*exec.ExitError); ok {
				if status, ok := exiterr.Sys().(syscall.WaitStatus); ok {
					switch status.ExitStatus() {
					case exitUpgrading:
						// Restart the monitor process to release the .old
						// binary as part of the upgrade process.
						l.Infoln("Restarting monitor...")
						os.Setenv("STNORESTART", "")

						if err = restartMonitor(args); err != nil {
							l.Warnln("Restart:", err)
						}

						return
					}
				}
			}
		}

		l.Infoln("Syncthing exited:", err)


		//TODO:[sunylat] 这个休眠时间决定了监控的间隔时间!
		time.Sleep(1 * time.Second)
	}
}

(2),main.go中的"syncthingMain(options)"函数:

/*
TODO:[sunylat] 提供真正服务的syncthing入口函数
*/
func syncthingMain(runtimeOptions RuntimeOptions) {

	//l.Infoln("[syncthingMain] 这是我的测试信息!")

	setupSignalHandling()

	// Create a main service manager. We'll add things to this as we go along.
	// We want any logging it does to go through our log system.

	//TODO:[sunylat] [begin] supervisor创建  ----------------------
	mainService := suture.New("main", suture.Spec{
		Log: func(line string) {
			l.Debugln(line)
		},
	})
	mainService.ServeBackground()

	// Set a log prefix similar to the ID we will have later on, or early log
	// lines look ugly.
	//l.SetPrefix("[start] ")


	//TODO:[sunylat] 我修改的启动时候显示的log头,加入了[syncthingMain]
	l.SetPrefix("[syncthingMain][start] ")

	if runtimeOptions.auditEnabled {
		startAuditing(mainService, runtimeOptions.auditFile)
	}

	if runtimeOptions.verbose {
		mainService.Add(newVerboseService())
	}

	errors := logger.NewRecorder(l, logger.LevelWarn, maxSystemErrors, 0)
	systemLog := logger.NewRecorder(l, logger.LevelDebug, maxSystemLog, initialSystemLog)

	// Event subscription for the API; must start early to catch the early
	// events. The LocalChangeDetected event might overwhelm the event
	// receiver in some situations so we will not subscribe to it here.
	apiSub := events.NewBufferedSubscription(events.Default.Subscribe(events.AllEvents&^events.LocalChangeDetected&^events.RemoteChangeDetected), 1000)
	diskSub := events.NewBufferedSubscription(events.Default.Subscribe(events.LocalChangeDetected|events.RemoteChangeDetected), 1000)

	if len(os.Getenv("GOMAXPROCS")) == 0 {
		runtime.GOMAXPROCS(runtime.NumCPU())
	}

	// Attempt to increase the limit on number of open files to the maximum
	// allowed, in case we have many peers. We don't really care enough to
	// report the error if there is one.
	osutil.MaximizeOpenFileLimit()

	// Ensure that that we have a certificate and key.
	cert, err := tls.LoadX509KeyPair(locations[locCertFile], locations[locKeyFile])
	if err != nil {
		l.Infof("Generating ECDSA key and certificate for %s...", tlsDefaultCommonName)
		cert, err = tlsutil.NewCertificate(locations[locCertFile], locations[locKeyFile], tlsDefaultCommonName, bepRSABits)
		if err != nil {
			l.Fatalln(err)
		}
	}

	myID = protocol.NewDeviceID(cert.Certificate[0])

	//这里把调试信息头设置为当前设备ID的一部分
	//l.SetPrefix(fmt.Sprintf("[%s] ", myID.String()[:5]))


	//TODO:[sunylat] 修改log头为[syncthingMain]
	l.SetPrefix("[syncthingMain] ")

	//打印syncthing长版本
	l.Infoln(LongVersion)
	l.Infoln("My ID:", myID)

	sha256.SelectAlgo()
	sha256.Report()
	perfWithWeakHash := cpuBench(3, 150*time.Millisecond, true)
	l.Infof("Hashing performance with weak hash is %.02f MB/s", perfWithWeakHash)
	perfWithoutWeakHash := cpuBench(3, 150*time.Millisecond, false)
	l.Infof("Hashing performance without weak hash is %.02f MB/s", perfWithoutWeakHash)

	// Emit the Starting event, now that we know who we are.

	events.Default.Log(events.Starting, map[string]string{
		"home": baseDirs["config"],
		"myID": myID.String(),
	})

	cfg := loadOrCreateConfig()

	if err := checkShortIDs(cfg); err != nil {
		l.Fatalln("Short device IDs are in conflict. Unlucky!
  Regenerate the device ID of one of the following:
  ", err)
	}

	if len(runtimeOptions.profiler) > 0 {
		go func() {
			l.Debugln("Starting profiler on", runtimeOptions.profiler)
			runtime.SetBlockProfileRate(1)
			err := http.ListenAndServe(runtimeOptions.profiler, nil)
			if err != nil {
				l.Fatalln(err)
			}
		}()
	}

	// The TLS configuration is used for both the listening socket and outgoing
	// connections.

	tlsCfg := &tls.Config{
		Certificates:           []tls.Certificate{cert},
		NextProtos:             []string{bepProtocolName},
		ClientAuth:             tls.RequestClientCert,
		SessionTicketsDisabled: true,
		InsecureSkipVerify:     true,
		MinVersion:             tls.VersionTLS12,
		CipherSuites: []uint16{
			0xCCA8, // TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305, Go 1.8
			0xCCA9, // TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305, Go 1.8
			tls.TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384,
			tls.TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384,
			tls.TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256,
			tls.TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256,
			tls.TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,
			tls.TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA,
			tls.TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA,
			tls.TLS_ECDHE_ECDSA_WITH_AES_256_CBC_SHA,
		},
	}

	opts := cfg.Options()

	if !opts.SymlinksEnabled {
		symlinks.Supported = false
	}

	if opts.WeakHashSelectionMethod == config.WeakHashAuto {
		if perfWithoutWeakHash*0.8 > perfWithWeakHash {
			l.Infof("Weak hash disabled, as it has an unacceptable performance impact.")
			weakhash.Enabled = false
		} else {
			l.Infof("Weak hash enabled, as it has an acceptable performance impact.")
			weakhash.Enabled = true
		}
	} else if opts.WeakHashSelectionMethod == config.WeakHashNever {
		l.Infof("Disabling weak hash")
		weakhash.Enabled = false
	} else if opts.WeakHashSelectionMethod == config.WeakHashAlways {
		l.Infof("Enabling weak hash")
		weakhash.Enabled = true
	}

	if (opts.MaxRecvKbps > 0 || opts.MaxSendKbps > 0) && !opts.LimitBandwidthInLan {
		lans, _ = osutil.GetLans()
		for _, lan := range opts.AlwaysLocalNets {
			_, ipnet, err := net.ParseCIDR(lan)
			if err != nil {
				l.Infoln("Network", lan, "is malformed:", err)
				continue
			}
			lans = append(lans, ipnet)
		}

		networks := make([]string, len(lans))
		for i, lan := range lans {
			networks[i] = lan.String()
		}
		l.Infoln("Local networks:", strings.Join(networks, ", "))
	}

	/*
	TODO:[sunylat] 打开本地数据库操作
	 */
	//TODO:[sunylat] 获取本地数据库全路径名称
	dbFile := locations[locDatabase]
	//TODO:[sunylat] 打开本地数据库
	ldb, err := db.Open(dbFile)

	if err != nil {
		l.Fatalln("Cannot open database:", err, "- Is another copy of Syncthing already running?")
	}

	if runtimeOptions.resetDeltaIdxs {
		l.Infoln("Reinitializing delta index IDs")
		ldb.DropDeltaIndexIDs()
	}

	protectedFiles := []string{
		locations[locDatabase],
		locations[locConfigFile],
		locations[locCertFile],
		locations[locKeyFile],
	}

	// Remove database entries for folders that no longer exist in the config
	folders := cfg.Folders()
	for _, folder := range ldb.ListFolders() {
		if _, ok := folders[folder]; !ok {
			l.Infof("Cleaning data for dropped folder %q", folder)
			db.DropFolder(ldb, folder)
		}
	}

	if cfg.RawCopy().OriginalVersion == 15 {
		// The config version 15->16 migration is about handling ignores and
		// delta indexes and requires that we drop existing indexes that
		// have been incorrectly ignore filtered.
		ldb.DropDeltaIndexIDs()
	}

	//TODO:[sunylat] 创建模型
	m := model.NewModel(cfg, myID, myDeviceName(cfg), "syncthing", Version, ldb, protectedFiles)

	if t := os.Getenv("STDEADLOCKTIMEOUT"); len(t) > 0 {
		it, err := strconv.Atoi(t)
		if err == nil {
			m.StartDeadlockDetector(time.Duration(it) * time.Second)
		}
	} else if !IsRelease || IsBeta {
		m.StartDeadlockDetector(20 * time.Minute)
	}

	if runtimeOptions.unpaused {
		setPauseState(cfg, false)
	} else if runtimeOptions.paused {
		setPauseState(cfg, true)
	}

	// Add and start folders
	for _, folderCfg := range cfg.Folders() {
		if folderCfg.Paused {
			continue
		}
		m.AddFolder(folderCfg)
		m.StartFolder(folderCfg.ID)
	}

	mainService.Add(m)

	// Start discovery

	cachedDiscovery := discover.NewCachingMux()
	mainService.Add(cachedDiscovery)

	// Start connection management

	connectionsService := connections.NewService(cfg, myID, m, tlsCfg, cachedDiscovery, bepProtocolName, tlsDefaultCommonName, lans)
	mainService.Add(connectionsService)

	if cfg.Options().GlobalAnnEnabled {
		for _, srv := range cfg.GlobalDiscoveryServers() {
			l.Infoln("Using discovery server", srv)
			gd, err := discover.NewGlobal(srv, cert, connectionsService)
			if err != nil {
				l.Warnln("Global discovery:", err)
				continue
			}

			// Each global discovery server gets its results cached for five
			// minutes, and is not asked again for a minute when it's returned
			// unsuccessfully.
			cachedDiscovery.Add(gd, 5*time.Minute, time.Minute, globalDiscoveryPriority)
		}
	}

	if cfg.Options().LocalAnnEnabled {
		// v4 broadcasts
		bcd, err := discover.NewLocal(myID, fmt.Sprintf(":%d", cfg.Options().LocalAnnPort), connectionsService)
		if err != nil {
			l.Warnln("IPv4 local discovery:", err)
		} else {
			cachedDiscovery.Add(bcd, 0, 0, ipv4LocalDiscoveryPriority)
		}
		// v6 multicasts
		mcd, err := discover.NewLocal(myID, cfg.Options().LocalAnnMCAddr, connectionsService)
		if err != nil {
			l.Warnln("IPv6 local discovery:", err)
		} else {
			cachedDiscovery.Add(mcd, 0, 0, ipv6LocalDiscoveryPriority)
		}
	}

	// GUI

	setupGUI(mainService, cfg, m, apiSub, diskSub, cachedDiscovery, connectionsService, errors, systemLog, runtimeOptions)

	if runtimeOptions.cpuProfile {
		f, err := os.Create(fmt.Sprintf("cpu-%d.pprof", os.Getpid()))
		if err != nil {
			log.Fatal(err)
		}
		pprof.StartCPUProfile(f)
	}

	for _, device := range cfg.Devices() {
		if len(device.Name) > 0 {
			l.Infof("Device %s is %q at %v", device.DeviceID, device.Name, device.Addresses)
		}
	}

	// Candidate builds always run with usage reporting.

	if IsCandidate {
		l.Infoln("Anonymous usage reporting is always enabled for candidate releases.")
		opts.URAccepted = usageReportVersion
		// Unique ID will be set and config saved below if necessary.
	}

	if opts.URAccepted > 0 && opts.URAccepted < usageReportVersion {
		l.Infoln("Anonymous usage report has changed; revoking acceptance")
		opts.URAccepted = 0
		opts.URUniqueID = ""
		cfg.SetOptions(opts)
	}

	if opts.URAccepted >= usageReportVersion && opts.URUniqueID == "" {
		// Generate and save a new unique ID if it is missing.
		opts.URUniqueID = rand.String(8)
		cfg.SetOptions(opts)
		cfg.Save()
	}

	// The usageReportingManager registers itself to listen to configuration
	// changes, and there's nothing more we need to tell it from the outside.
	// Hence we don't keep the returned pointer.
	newUsageReportingManager(cfg, m)

	if opts.RestartOnWakeup {
		go standbyMonitor()
	}

	// Candidate builds should auto upgrade. Make sure the option is set,
	// unless we are in a build where it's disabled or the STNOUPGRADE
	// environment variable is set.

	if IsCandidate && !upgrade.DisabledByCompilation && !noUpgradeFromEnv {
		l.Infoln("Automatic upgrade is always enabled for candidate releases.")
		if opts.AutoUpgradeIntervalH == 0 || opts.AutoUpgradeIntervalH > 24 {
			opts.AutoUpgradeIntervalH = 12
		}
		// We don't tweak the user's choice of upgrading to pre-releases or
		// not, as otherwise they cannot step off the candidate channel.
	}

	if opts.AutoUpgradeIntervalH > 0 {
		if noUpgradeFromEnv {
			l.Infof("No automatic upgrades; STNOUPGRADE environment variable defined.")
		} else {
			go autoUpgrade(cfg)
		}
	}

	events.Default.Log(events.StartupComplete, map[string]string{
		"myID": myID.String(),
	})

	cleanConfigDirectory()

	code := <-stop

	//TODO:[sunylat] [end] supervisor结束  ----------------------
	mainService.Stop()

	l.Infoln("Exiting")

	if runtimeOptions.cpuProfile {
		pprof.StopCPUProfile()
	}

	os.Exit(code)
}

提示:在"syncthingMain(options)"函数使用了第三方的库,使用的代码:

	mainService := suture.New("main", suture.Spec{
		Log: func(line string) {
			l.Debugln(line)
		},
	})
	mainService.ServeBackground()

第三方库的官方地址:

https://github.com/thejerf/suture

这是用GO语言写的Erlang语言中的supervisor trees功能,应该是很有用,值得我们关注!

Erlang supervisor trees 中文参考:

http://diaocow.iteye.com/blog/1762895

  

 

原文地址:https://www.cnblogs.com/sunylat/p/6389339.html