flink-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Till Rohrmann (JIRA)" <j...@apache.org>
Subject [jira] [Created] (FLINK-5836) Race condition between slot offering and task deployment
Date Fri, 17 Feb 2017 13:39:41 GMT
Till Rohrmann created FLINK-5836:

             Summary: Race condition between slot offering and task deployment
                 Key: FLINK-5836
                 URL: https://issues.apache.org/jira/browse/FLINK-5836
             Project: Flink
          Issue Type: Bug
          Components: Distributed Coordination
    Affects Versions: 1.3.0
            Reporter: Till Rohrmann

The Flip-6 code has a race condition when offering slots to a {{JobManager}} which directly
deploys tasks to the offered slots. In such a situation it is possible that the deploy call
overtakes the acknowledge message for the slot offering. As a result, the slots are not marked
yet as active and the deployment will fail.

I propose to fix this problem by first activating all offered slots before sending the slot
offer message to the {{JobManager}}. Consequently, we'll deactivate and free slots which haven't
been accepted by the {{JobManager}} once we've received the offering acknowledge message.

This message was sent by Atlassian JIRA

View raw message