[
https://issues.apache.org/jira/browse/NIFI-710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14980669#comment-14980669
]
Oleg Zhurakousky commented on NIFI-710:
---------------------------------------
I think we can do better. Instead of delegating the decision to the end user we
should simply ask hadoop for a versions (e.g., 'hadoop version') based on the
current configuration and use that to do a look up of all the required JARs. In
other words we would need some cache of JARs related to all versions of Hadoop
we choose to support.
> Enumerate available Hadoop library versions for GetHDFS and PutHDFS
> -------------------------------------------------------------------
>
> Key: NIFI-710
> URL: https://issues.apache.org/jira/browse/NIFI-710
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Extensions
> Environment: Unix, Hadoop
> Reporter: Nathan Gough
> Priority: Minor
> Labels: hadoop, hdfs, library
> Fix For: 0.4.0
>
>
> As far as I'm aware, it is only possible to use a single Hadoop library
> version per NiFi instance/cluster.
> Would it be possible in some way to enumerate the available Hadoop library
> versions in the /lib directory and give the user the option to select which
> version of Hadoop cluster they are writing to or reading from per HDFS
> processor? The intent would be to allow a single NiFi instance/cluster to
> write to multiple HDFS clusters which may be running on different versions
> eg. QA and Prod. This eliminates the need for having multiple HDFS feeder
> NiFis only because the Hadoop versions are different.
> I would try and figure this out my self however I don't know a whole lot
> about class loading in NiFi or generally how I would start this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
