GitHub user vjagadish opened a pull request: https://github.com/apache/samza/pull/488
SAMZA-1687: Prioritize preferred host requests over ANY-HOST requests Working on a documentation that describes this better, but a TL;DR summary is that we should prioritize preferred-host requests over ANY_HOST requests. Yarn enforces these two checks: 1. ANY_HOST requests should always be made with relax-locality = true 2. A request with relax-locality = false should not be in the same priority as another with relax-locality = true Since the Samza AM makes preferred-host requests with relax-locality = false, it follows that ANY_HOST requests should specify a different priority-level. We can safely set priority of preferred-host requests to be higher than any-host requests since data-locality is critical. You can merge this pull request into a Git repository by running: $ git pull https://github.com/vjagadish1989/samza priority-host-affinity Alternatively you can review and apply these changes as the patch at: https://github.com/apache/samza/pull/488.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #488 ---- commit 2b640f1ba6813f1b193e9433801ef3f5cae85551 Author: Jagadish <jvenkatraman@...> Date: 2018-04-21T02:49:54Z Update PMC membership on the Samza web-page commit 7c38f4990f024ebf7a5ea5815c1adf97ee0ff953 Author: Jagadish <jvenkatraman@...> Date: 2018-04-23T20:41:37Z Merge branch 'master' of https://github.com/apache/samza commit 32ea1c259e0eeb3fd94c84241b0012babe55eff8 Author: Jagadish <jvenkatraman@...> Date: 2018-04-25T18:17:31Z Prioritize preferred host requests over ANY_HOST requests ---- ---