You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@spark.apache.org by "Cheng Lian (JIRA)" <ji...@apache.org> on 2016/04/27 12:18:12 UTC

[jira] [Created] (SPARK-14953) LocalBackend should revive offers periodically

Cheng Lian created SPARK-14953:
----------------------------------

             Summary: LocalBackend should revive offers periodically
                 Key: SPARK-14953
                 URL: https://issues.apache.org/jira/browse/SPARK-14953
             Project: Spark
          Issue Type: Bug
          Components: Spark Core
            Reporter: Cheng Lian


{{LocalBackend}} only revives offers when tasks are submitted, succeed, or fail. This may lead to deadlock due to delayed scheduling. A case study is provided in [this PR comment|https://github.com/apache/spark/pull/12527#issuecomment-213034425].

Basically, a job may have a task is delayed to be scheduled due to locality mismatch. The default delay timeout is 3s. If all other tasks finish during this period, {{LocalBackend}} won't revive any offer after the timeout since no tasks are submitted, succeed or fail then. Thus, the delayed task will never be scheduled again and the job never completes.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org