Watchdog waits an entire period before first tick

[lahabana]

What happened?

From the code:

kuma/pkg/util/watchdog/watchdog.go

Lines 21 to 53 in ae945b0

	func (w *SimpleWatchdog) Start(stop <-chan struct{}) {
	ticker := w.NewTicker()
	defer ticker.Stop()
	for {
	ctx, cancel := context.WithCancel(context.Background())
	// cancel is called at the end of the loop
	go func() {
	select {
	case <-stop:
	cancel()
	case <-ctx.Done():
	}
	}()
	select {
	case <-ticker.C:
	select {
	case <-stop:
	default:
	if err := w.onTick(ctx); err != nil && !errors.Is(err, context.Canceled) {
	w.OnError(err)
	}
	}
	case <-stop:
	if w.OnStop != nil {
	w.OnStop()
	}
	// cancel will be called by the above goroutine
	return
	}
	cancel()
	}
	}

We probably should not wait an entire tick before triggering first call. This can delay things unecessarilly (for example sending the configuration).

[slonka]

triage: accepted