IMO the dashboard does not match the log error, or?
Spark master dashboard says (similar for memory as well):
Alive Workers: 2 Cores in use: 2 Total, 0 Used
17/07/21 20:48:45 WARN TaskSchedulerImpl: Initial job has not accepted any resources; check your cluster UI to ensure that workers are registered and have sufficient resources
Assuming there was e.g. not enough memory, how to determine that? Maybe a symptom, the workers get here and now "dissociated" but then reconnect.