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

Posted to yarn-issues@hadoop.apache.org by "Zhankun Tang (JIRA)" <ji...@apache.org> on 2019/01/02 06:22:00 UTC

[jira] [Created] (YARN-9167) [Submarine] Support fault tolerance when Tensorflow worker container fails

Zhankun Tang created YARN-9167:
----------------------------------

             Summary: [Submarine] Support fault tolerance when Tensorflow worker container fails
                 Key: YARN-9167
                 URL: https://issues.apache.org/jira/browse/YARN-9167
             Project: Hadoop YARN
          Issue Type: Sub-task
            Reporter: Zhankun Tang
            Assignee: Zhankun Tang


A long-running Tensorflow job needs to restart failed worker containers when something unexpected happens. Luckily that TF can restore checkpoints and continue training in a worker, a restart of the worker container seems enough.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

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