[
https://issues.apache.org/jira/browse/FLINK-1834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14483249#comment-14483249
]
Fabian Hueske commented on FLINK-1834:
--------------------------------------
I totally agree. We should get rid of this mandatory parameter.
We try to only set it if it is set by the user and let the HadoopFormat figure
out whether it needs it or not. If it does and it is not set, the job should
fail.
> Is mapred.output.dir conf parameter really required?
> ----------------------------------------------------
>
> Key: FLINK-1834
> URL: https://issues.apache.org/jira/browse/FLINK-1834
> Project: Flink
> Issue Type: Improvement
> Components: Hadoop Compatibility
> Affects Versions: 0.9
> Reporter: Flavio Pompermaier
> Priority: Minor
> Labels: hadoop
>
> At the moment if you don't set mapred.output.dir parameter when output to an
> hadoop sink Flink throws an error. Probably this parameter could be defaulted
> to some tmp dir + jobId (for each output sink).
> For example, when output to an HBase sink, the only things you'll find in the
> mapred.output.dir will be the SUCCESS files
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
