[ https://issues.apache.org/jira/browse/SPARK-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14744162#comment-14744162 ]
Tathagata Das commented on SPARK-7040: -------------------------------------- I am not sure how Direct API can be built for Flume as Flume does not have any offsets or sequence numbers (correct me if I am wrong about this) to refer to the exact ranges of records / events. I am closing this JIRA for now, please reopen it if you think this is still relevant. > Explore receiver-less DStream for Flume > --------------------------------------- > > Key: SPARK-7040 > URL: https://issues.apache.org/jira/browse/SPARK-7040 > Project: Spark > Issue Type: New Feature > Components: Streaming > Reporter: Hari Shreedharan > > I am thinking about repurposing the FlumePollingInputDStream to make it more > parallel and pull in data like the DirectKafkaDStream. Since Flume does not > have a unique way of identifying specific event offsets, this will not be > once-only. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org