Why would you have expected a problem? Even with a multi node manager setup, you can stop all the nodes and restart them later and your cluster will continue to work like it was never turned offf (give your tasks are deployed with the restart policy “any”)
With a single manager, the state is clear. There is nothing to synrononize between nodes and no consensus to be established for state changes. If the manager is down, the only thing that happens is, is that your workers are not controllable and neither new ressources can be deployed nor stopped tasks can be respawned until the manager is back online.