@Bilal Aslam​  might have a better phrasing of the question. What I mean is "how do I make sure my job does not start until all worker nodes in a Spark cluster are ready"