On a big cluster, it takes the server a long time to process the API request for certain operations, like "Stop All", "Start All", etc.
On a small cluster, the "background operations" popup appears within a few seconds to acknowledge to the user that the server is working on handling the request.
However, when the cluster gets bigger and the request API takes the server much longer (e.g., on a 1000-node cluster, it took ~35 seconds), there's no indication on the UI that the server received the request. This is not reassuring to the user and the user is tempted to keep re-triggering the action and overwhelming the server.
This problem is not just for "Start All", "Stop All". This problem actually happens for "Stop" and "Start" of a single service as well (and likely other places, like Hosts / Host Detail pages.) On the 1000-node cluster, this takes about 6-7 seconds, enough to make the user wonder if the request got to the server or not.