[jira] [Created] (AMBARI-4378) Rolling restart toleration count - support absolute values only

["Siddharth Wagle (JIRA)" <ji...@apache.org>]

Siddharth Wagle created AMBARI-4378:
---------------------------------------

             Summary: Rolling restart toleration count - support absolute values only
                 Key: AMBARI-4378
                 URL: https://issues.apache.org/jira/browse/AMBARI-4378
             Project: Ambari
          Issue Type: Task
          Components: controller
    Affects Versions: 1.5.0
            Reporter: Siddharth Wagle
            Assignee: Siddharth Wagle
             Fix For: 1.5.0
         Attachments: AMBARI-4378.patch

The "task_failure_tolerance" applies to percentage of tasks created thus far and for the present request, since we do not now how many tasks the proceeding request will create. Consider it as a cumulative count, so in the first batch, 1 out of 2 tasks failed, which is 50% failure and allowed tolerance is 25% so this will fail.

Different test cases:
Try 4 batches of 1 host each and make the 3rd batch fail, in this case failure when 3rd batch fails = 1/3 : 33.33% failure.
So if you set "task_failure_tolerance" to 40, 4th batch should start, and if you set it to 25, 4th batch will not start.

For a Rolling restart scenario and absolute count makes more sense.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)