[jira] [Created] (FLINK-1288) YARN ApplicationMaster sometimes fails to allocate the specified number of workers

["Robert Metzger (JIRA)" <ji...@apache.org>]

Robert Metzger created FLINK-1288:
-------------------------------------

             Summary: YARN ApplicationMaster sometimes fails to allocate the specified number of workers
                 Key: FLINK-1288
                 URL: https://issues.apache.org/jira/browse/FLINK-1288
             Project: Flink
          Issue Type: Bug
          Components: YARN Client
            Reporter: Robert Metzger
            Assignee: Robert Metzger


There seems to be a bug somewhere that Flink fails to allocate a full YARN cluster.

For example, my cluster has 42 nodes with 46.09 GB each (according to the NodeManager)
If my Flink client is requesting 41 + 1 (workers + master) containers (with 46GB each), I'm only getting 41 containers (instead of 42).

I've contacted the yarn-dev list already for help and I'm trying to figure out the cause of the issue.

As a simple workaround, users can just request one more container than they actually need.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)