Merge pull request #39105 from cpuguy83/optimize_test_daemon_startup

Optimize test daemon startup

cmd/dockerd/daemon.go

@@ -91,6 +91,8 @@ func (cli *DaemonCli) start(opts *daemonOptions) (err error) {
 		return err
 	}
 
+	logrus.Info("Starting up")
+
 	cli.configFile = &opts.configFile
 	cli.flags = opts.flags
 
@@ -266,6 +268,7 @@ func (cli *DaemonCli) start(opts *daemonOptions) (err error) {
 		return errors.Wrap(errAPI, "shutting down due to ServeAPI error")
 	}
 
+	logrus.Info("Daemon shutdown complete")
 	return nil
 }
 

cmd/dockerd/daemon_unix.go

@@ -20,6 +20,7 @@ import (
 	"github.com/docker/docker/pkg/homedir"
 	"github.com/docker/libnetwork/portallocator"
 	"github.com/pkg/errors"
+	"github.com/sirupsen/logrus"
 	"golang.org/x/sys/unix"
 )
 
@@ -152,6 +153,7 @@ func (cli *DaemonCli) initContainerD(ctx context.Context) (func(time.Duration) e
 			return nil, errors.Wrap(err, "could not determine whether the system containerd is running")
 		}
 		if !ok {
+			logrus.Debug("Containerd not running, starting daemon managed containerd")
 			opts, err := cli.getContainerdDaemonOpts()
 			if err != nil {
 				return nil, errors.Wrap(err, "failed to generate containerd options")
@@ -161,6 +163,7 @@ func (cli *DaemonCli) initContainerD(ctx context.Context) (func(time.Duration) e
 			if err != nil {
 				return nil, errors.Wrap(err, "failed to start containerd")
 			}
+			logrus.Debug("Started daemon managed containerd")
 			cli.Config.ContainerdAddr = r.Address()
 
 			// Try to wait for containerd to shutdown

internal/test/daemon/daemon.go

@@ -290,27 +290,10 @@ func (d *Daemon) StartWithLogFile(out *os.File, providedArgs ...string) error {
 
 	d.Wait = wait
 
-	ticker := time.NewTicker(500 * time.Millisecond)
-	defer ticker.Stop()
-	tick := ticker.C
-
-	// make sure daemon is ready to receive requests
-	startTime := time.Now().Unix()
-	for {
-		d.log.Logf("[%s] waiting for daemon to start", d.id)
-		if time.Now().Unix()-startTime > 5 {
-			// After 5 seconds, give up
-			return errors.Errorf("[%s] Daemon exited and never started", d.id)
-		}
-		select {
-		case <-time.After(2 * time.Second):
-			return errors.Errorf("[%s] timeout: daemon does not respond", d.id)
-		case <-tick:
 	clientConfig, err := d.getClientConfig()
 	if err != nil {
 		return err
 	}
-
 	client := &http.Client{
 		Transport: clientConfig.transport,
 	}
@@ -321,10 +304,30 @@ func (d *Daemon) StartWithLogFile(out *os.File, providedArgs ...string) error {
 	}
 	req.URL.Host = clientConfig.addr
 	req.URL.Scheme = clientConfig.scheme
-			resp, err := client.Do(req)
+
+	ticker := time.NewTicker(500 * time.Millisecond)
+	defer ticker.Stop()
+	tick := ticker.C
+
+	timeout := time.NewTimer(60 * time.Second) // timeout for the whole loop
+	defer timeout.Stop()
+
+	// make sure daemon is ready to receive requests
+	for {
+		d.log.Logf("[%s] waiting for daemon to start", d.id)
+
+		select {
+		case <-timeout.C:
+			return errors.Errorf("[%s] Daemon exited and never started", d.id)
+		case <-tick:
+			ctx, cancel := context.WithTimeout(context.TODO(), 2*time.Second)
+			resp, err := client.Do(req.WithContext(ctx))
+			cancel()
 			if err != nil {
+				d.log.Logf("[%s] error pinging daemon on start: %v", d.id, err)
 				continue
 			}
+
 			resp.Body.Close()
 			if resp.StatusCode != http.StatusOK {
 				d.log.Logf("[%s] received status != 200 OK: %s\n", d.id, resp.Status)
@@ -420,12 +423,16 @@ func (d *Daemon) Stop(t testingT) {
 // If it timeouts, a SIGKILL is sent.
 // Stop will not delete the daemon directory. If a purged daemon is needed,
 // instantiate a new one with NewDaemon.
-func (d *Daemon) StopWithError() error {
+func (d *Daemon) StopWithError() (err error) {
 	if d.cmd == nil || d.Wait == nil {
 		return errDaemonNotStarted
 	}
-
 	defer func() {
+		if err == nil {
+			d.log.Logf("[%s] Daemon stopped", d.id)
+		} else {
+			d.log.Logf("[%s] Error when stopping daemon: %v", d.id, err)
+		}
 		d.logFile.Close()
 		d.cmd = nil
 	}()
@@ -435,12 +442,15 @@ func (d *Daemon) StopWithError() error {
 	defer ticker.Stop()
 	tick := ticker.C
 
+	d.log.Logf("[%s] Stopping daemon", d.id)
+
 	if err := d.cmd.Process.Signal(os.Interrupt); err != nil {
 		if strings.Contains(err.Error(), "os: process already finished") {
 			return errDaemonNotStarted
 		}
 		return errors.Errorf("could not send signal: %v", err)
 	}
+
 out1:
 	for {
 		select {