Currently there is an exponential backoff during retries and all masters will be tested but there is no limit on the number of retries so if one instance of rhythm has connection issues to Mesos cluster and it's elected as leader then it won't give up - it won't give a chance to other rhythm instances to give a try.
Currently there is an exponential backoff during retries and all masters will be tested but there is no limit on the number of retries so if one instance of
rhythmhas connection issues to Mesos cluster and it's elected as leader then it won't give up - it won't give a chance to otherrhythminstances to give a try.