commit | 3fe637eb573c0d8aa85420c5d6101efd6aef13e4 | [log] [tgz] |
---|---|---|
author | Zhao, Qingwen <qingwzhao@apache.org> | Wed Mar 15 16:34:10 2017 +0800 |
committer | Zhao, Qingwen <qingwzhao@apache.org> | Wed Mar 15 16:34:10 2017 +0800 |
tree | 0b79ea265fa19f608253720d3050762f4e9a90e1 | |
parent | 93f83f4a975dd3303f776c4fc9b3da8a53384aa3 [diff] |
[EAGLE-946] Refactor MRRunningJobApp & HadoopQueueApp https://issues.apache.org/jira/browse/EAGLE-946 Since upgrading to HDP 2.7, the api of fetching the running app list in one request become very heavy. To tackle this issue, a new algorithm is introduced: `if requests# <=1, fetch all running job list in one request; otherwise, divide last 6 hours into requests# slots, and fetch running jobs in each slot one by one` Author: Zhao, Qingwen <qingwzhao@apache.org> Closes #869 from qingwen220/EAGLE-946.