GitHub user shivaram opened a pull request: https://github.com/apache/spark/pull/1697
[SPARK-2774 - Set preferred locations for reduce tasks Motivation for the change is in JIRA. There are a couple of things that I would like feedback about 1. Should we sort the map outputs by size for every task -- This could be expensive if we have a large number of map outputs. 2. The number of preferred locations to use. Technically we could set this to a larger number, but I am not sure how it will affect the locality / delay scheduling in TaskSetManager. cc @rxin You can merge this pull request into a Git repository by running: $ git pull https://github.com/shivaram/spark-1 reducer-locality Alternatively you can review and apply these changes as the patch at: https://github.com/apache/spark/pull/1697.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 #1697 ---- commit 3fe76f7a22c663ae1f28585b6e2c15696f46e0d9 Author: Shivaram Venkataraman <shiva...@cs.berkeley.edu> Date: 2014-07-31T04:15:53Z Set preferred locations for reduce tasks commit f8390ddb0e9f50ec734735ea4ec80f5af419d359 Author: Shivaram Venkataraman <shiva...@cs.berkeley.edu> Date: 2014-07-31T04:55:12Z Amortize traversing array by doing it once per RDD. commit 6782dea8aa36628b5319191e6f52a8f62fa6f98c Author: Shivaram Venkataraman <shiva...@cs.berkeley.edu> Date: 2014-07-31T18:04:00Z Merge branch 'master' of https://github.com/apache/spark into reducer-locality commit 3666ff57fd1d25761edaadbf34d8287ff6df25f8 Author: Shivaram Venkataraman <shiva...@cs.berkeley.edu> Date: 2014-07-31T18:29:57Z Add a unit test that checks for reducer locality ---- --- 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. ---