Hi @evgeniya.bell - every handler runs on each eventing node, as we shard the data across the eventing nodes (rather than assigning some handlers to some eventing nodes, and having all nodes receive all data). So the UI warning looks at number of cores and number of handlers only, and not the number of nodes.
The purpose of the warning is to basically draw attention to the need to do sizing exercise. There is no hard limit that number of workers must be less than number of cores. When running handlers that have high latency (ex: making a curl call to outside world), it is common to have many more workers than number of cores.
As we don’t know beforehand what the handler code is doing, or how much mutation workload is expected, the warning uses this rudimentary approach to draw attention to sizing. We will change the text of the warning to be less confusing (and merely state that a sizing exercise should be performed) to avoid further ambiguity, MB-33624.