Github user sryza commented on the pull request: https://github.com/apache/spark/pull/634#issuecomment-42146849 A little more detail on the JIRA, but, to summarize, the sleep is to wait for executors to connect. What's confusing to me is that Spark standalone doesn't have a similar sleep. @mridulm is the worry that only some executors would register and that tasks would then unnecessarily get scheduled on nodes where the data isn't local? Have you seen this negatively impact performance? If that's the case, I think the best thing would be to wait until some number of executors have registered.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---