It is a nice feature that in case one render slave isnt able to a render its bucket that it is flagged as “not responding”. However is there a control to set the timer to a specific amount. A workaround would be to decrease the bucket size. I havent found any information on this.
Also, the communication that happens to find out if a node is responding is separate from the rendering; so ideally even though the buckets are big, the render slave should still respond to ping messages from the main machine to let it know that it is alive.