Make Echo method to be a really timebound call

The client's Echo method invokes o.rpcClient.CallWithContext to send echo
to server which again invokes c.Go which writes request into channel buffer
synchronizely which is not covered with context timeout. So timeout never
triggered and inactivity check is indefinitely waiting for the echo reply.
Hence this commit makes Echo method itslef to honor the context timeout
at any case.

Signed-off-by: Periyasamy Palanisamy <[email protected]>

client/client.go

@@ -1080,16 +1080,27 @@ func (o *ovsdbClient) Echo(ctx context.Context) error {
 	if o.rpcClient == nil {
 		return ErrNotConnected
 	}
-	err := o.rpcClient.CallWithContext(ctx, "echo", args, &reply)
-	if err != nil {
-		if err == rpc2.ErrShutdown {
-			return ErrNotConnected
+	errCh := make(chan error)
+	go func() {
+		err := o.rpcClient.CallWithContext(ctx, "echo", args, &reply)
+		if err != nil {
+			if err == rpc2.ErrShutdown {
+				errCh <- ErrNotConnected
+				return
+			}
 		}
+		if !reflect.DeepEqual(args, reply) {
+			errCh <- fmt.Errorf("incorrect server response: %v, %v", args, reply)
+			return
+		}
+		errCh <- nil
+	}()
+	select {
+	case err := <-errCh:
+		return err
+	case <-ctx.Done():
+		return ctx.Err()
 	}
-	if !reflect.DeepEqual(args, reply) {
-		return fmt.Errorf("incorrect server response: %v, %v", args, reply)
-	}
-	return nil
 }
 
 // watchForLeaderChange will trigger a reconnect if the connected endpoint