Github user mridulm commented on the pull request: https://github.com/apache/spark/pull/663#issuecomment-42356011 That sounds reasonable approach - will know more when PR is updated. The initial idea behind separate Thread for each was exactly that - we expected low hundreds as upper limit (at that time); we needed to send request to RM asap while not being sure how long the rpc can take. This is, clearly, suboptimal leading to unbounded increase in threads during initial allocation burst.
--- 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. ---