client/v3: Add backoff before retry when watch stream returns unavail…

…able

The client retries connection without backoff when the server is gone
after the watch stream is established. This results in high CPU usage
in the client process. This change introduces backoff when the stream is
failed and unavailable.

Signed-off-by: Hisanobu Tomari <[email protected]>

client/v3/watch.go

@@ -540,6 +540,7 @@ func (w *watchGrpcStream) run() {
 	cancelSet := make(map[int64]struct{})
 
 	var cur *pb.WatchResponse
+	backoff := time.Millisecond
 	for {
 		select {
 		// Watch() requested
@@ -676,6 +677,7 @@ func (w *watchGrpcStream) run() {
 				closeErr = err
 				return
 			}
+			backoff = w.backoffIfUnavailable(backoff, err)
 			if wc, closeErr = w.newWatchClient(); closeErr != nil {
 				return
 			}
@@ -995,6 +997,21 @@ func (w *watchGrpcStream) joinSubstreams() {
 
 var maxBackoff = 100 * time.Millisecond
 
+func (w *watchGrpcStream) backoffIfUnavailable(backoff time.Duration, err error) time.Duration {
+	if isUnavailableErr(w.ctx, err) {
+		// retry, but backoff
+		if backoff < maxBackoff {
+			// 25% backoff factor
+			backoff = backoff + backoff/4
+			if backoff > maxBackoff {
+				backoff = maxBackoff
+			}
+		}
+		time.Sleep(backoff)
+	}
+	return backoff
+}
+
 // openWatchClient retries opening a watch client until success or halt.
 // manually retry in case "ws==nil && err==nil"
 // TODO: remove FailFast=false
@@ -1015,17 +1032,7 @@ func (w *watchGrpcStream) openWatchClient() (ws pb.Watch_WatchClient, err error)
 		if isHaltErr(w.ctx, err) {
 			return nil, v3rpc.Error(err)
 		}
-		if isUnavailableErr(w.ctx, err) {
-			// retry, but backoff
-			if backoff < maxBackoff {
-				// 25% backoff factor
-				backoff = backoff + backoff/4
-				if backoff > maxBackoff {
-					backoff = maxBackoff
-				}
-			}
-			time.Sleep(backoff)
-		}
+		backoff = w.backoffIfUnavailable(backoff, err)
 	}
 	return ws, nil
 }