[jira] [Commented] (AIRFLOW-1718) Increase num_retries polling value on Dataproc hook

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AIRFLOW-1718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16209792#comment-16209792 ] 

ASF subversion and git services commented on AIRFLOW-1718:
----------------------------------------------------------

Commit 6078e753aac35aa4f5971a719d9f736c35396770 in incubator-airflow's branch refs/heads/master from [~cjqian]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-airflow.git;h=6078e75 ]

[AIRFLOW-1718] Set num_retries on Dataproc job request execution

Closes #2696 from cjqian/1718


> Increase num_retries polling value on Dataproc hook
> ---------------------------------------------------
>
>                 Key: AIRFLOW-1718
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-1718
>             Project: Apache Airflow
>          Issue Type: Bug
>            Reporter: Crystal Qian
>            Assignee: Crystal Qian
>            Priority: Minor
>   Original Estimate: 24h
>  Remaining Estimate: 24h
>
> Currently, `num_retries = 0` when execute() is called (https://google.github.io/google-api-python-client/docs/epy/googleapiclient.http.HttpRequest-class.html#execute), which causes intermittent 500 errors (https://stackoverflow.com/questions/46522261/deadline-exceeded-when-airflow-runs-spark-jobs). We should increase this to allow retries for internal Dataproc queries to other services in the short-term; also seeing if the `num_retries` count can be increased at the _google-api-python-client_ level in the long-term.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)