[
https://issues.apache.org/jira/browse/CASSANDRA-7827?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14162014#comment-14162014
]
Paul Pak commented on CASSANDRA-7827:
-------------------------------------
[~pkolaczk] I'm not exactly sure why. Perhaps it's because the namedOutput name
is potentially used as part of the HDFS file path and there are
restrictions/conventions around that? Either way,
MultipleOutputs.checkNamedOutputName() -> MultipleOutputs.checkTokenName()
restricts the name to only [A-Za-z0-9].
> Work around for output name restriction when using MultipleOutputs with
> CqlBulkOutputFormat
> -------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-7827
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7827
> Project: Cassandra
> Issue Type: Improvement
> Components: Hadoop
> Reporter: Paul Pak
> Assignee: Paul Pak
> Priority: Minor
> Labels: cql3, hadoop
> Attachments: trunk-7827-v1.txt
>
>
> When using MultipleOutputs with CqlBulkOutputFormat, the column family names
> to output to are restricted to only alphanumeric characters due to the logic
> found in MultipleOutputs.checkNamedOutputName(). This will provide a way to
> alias any column family name to a MultipleOutputs compatible output name, so
> that column family names won't be artificially restricted when using
> MultipleOutputs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
