You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by "Junyoung Park (JIRA)" <ji...@apache.org> on 2017/12/18 11:42:00 UTC

[jira] [Commented] (AIRFLOW-1936) EmrCreateJobFlowOperator is unable to launch the aws emr cluster

    [ https://issues.apache.org/jira/browse/AIRFLOW-1936?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16294863#comment-16294863 ] 

Junyoung Park commented on AIRFLOW-1936:
----------------------------------------

It looks weird, cause it worked well on my EMR cluster. (v5.7 ~ 5.10)
Have you tried the following methods?

1. Go to Airflow - Admin - Connection sections.
2. Fix extra value in 'aws_default' Conn Id. (JSON type)
3. Fix extra value in 'emr_default' Conn Id. (JSON type)
4. run DAG with no job_flow_overrides.

Or check that the job_flow_overrides value is valid.
If the cluster creation fails, Airflow prints an error message.

> EmrCreateJobFlowOperator is unable to launch the aws emr cluster
> ----------------------------------------------------------------
>
>                 Key: AIRFLOW-1936
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-1936
>             Project: Apache Airflow
>          Issue Type: Bug
>          Components: aws, boto3, DAG, hooks, operators
>    Affects Versions: Airflow 1.8
>         Environment: ubuntu 16.04 , python 2.7, boto3
>            Reporter: Vikram Fugro
>              Labels: newbie
>         Attachments: dag_logs.txt, emr_hook_with_hardcoded_values.py, my_emr_boto_script(works).py, my_emr_dag(doesNotWork).py
>
>
> The EmrCreateJobFlowOperator  operator is unable to create the emr cluster, although it returns  success. The most strange thing that I see in the logs is that it also returns the jobflow-Id but I see no cluster starting up in the aws emr console. If I try to launch the emr with my boto3 script , it works. I even  hardcoded the arguments (taken from my boto3 script as is)  in self.get_conn().run_job_flow(), in contrib/hooks/emr_hook.py , but no luck. 
> I am running airflow 1.8, localexecutor with metadb postgres. I have tried both ways ; setting up a schedule and also by  command 'airflow run  -f emr_job_flow_manual_steps_dag2 create_job_flow 2017-12-18' 
> I am attaching the logs of my dag,  my dag itself , my boto3 script and  emr_hook.py(changed only for testing purpose).



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