etcdctl: cluster-health supports forever flag

cluster-health command supports checking the cluster health
forever.

etcdctl/command/cluster_health.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"net/http"
 	"os"
+	"os/signal"
 	"sort"
 	"time"
 
@@ -17,33 +18,52 @@ func NewClusterHealthCommand() cli.Command {
 	return cli.Command{
 		Name:  "cluster-health",
 		Usage: "check the health of the etcd cluster",
-		Flags:  []cli.Flag{},
+		Flags: []cli.Flag{
+			cli.BoolFlag{Name: "forever", Usage: "forever check the health every 10 second until CTRL+C"},
+		},
 		Action: handleClusterHealth,
 	}
 }
 
 func handleClusterHealth(c *cli.Context) {
+	forever := c.Bool("forever")
+	if forever {
+		sigch := make(chan os.Signal, 1)
+		signal.Notify(sigch, os.Interrupt)
+
+		go func() {
+			<-sigch
+			os.Exit(0)
+		}()
+	}
+
 	tr, err := getTransport(c)
 	if err != nil {
 		handleError(ExitServerError, err)
 	}
 
+	// TODO: update members when forever is set.
 	mi := mustNewMembersAPI(c)
 	ms, err := mi.List(context.TODO())
 	if err != nil {
+		fmt.Println("cluster may be unhealthy: failed to list members")
 		handleError(ExitServerError, err)
 	}
-
 	cl := make([]string, 0)
 	for _, m := range ms {
 		cl = append(cl, m.ClientURLs...)
 	}
 
+	for {
 		// check the /health endpoint of all members first
 
 		ep, rs0, err := getLeaderStatus(tr, cl)
 		if err != nil {
 			fmt.Println("cluster may be unhealthy: failed to connect", cl)
+			if forever {
+				time.Sleep(10 * time.Second)
+				continue
+			}
 			os.Exit(1)
 		}
 
@@ -53,6 +73,10 @@ func handleClusterHealth(c *cli.Context) {
 		_, rs1, err := getLeaderStatus(tr, []string{ep})
 		if err != nil {
 			fmt.Println("cluster is unhealthy")
+			if forever {
+				time.Sleep(10 * time.Second)
+				continue
+			}
 			os.Exit(1)
 		}
 
@@ -68,6 +92,7 @@ func handleClusterHealth(c *cli.Context) {
 		for id, pr0 := range rs0.Progress {
 			pr1, ok := rs1.Progress[id]
 			if !ok {
+				// TODO: forever should handle configuration change.
 				fmt.Println("Cluster configuration changed during health checking. Please retry.")
 				os.Exit(1)
 			}
@@ -82,7 +107,13 @@ func handleClusterHealth(c *cli.Context) {
 		for _, p := range prints {
 			fmt.Print(p)
 		}
-	os.Exit(0)
+
+		if !forever {
+			return
+		}
+
+		time.Sleep(10 * time.Second)
+	}
 }
 
 type raftStatus struct {