spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matei Zaharia <matei.zaha...@gmail.com>
Subject Re: Building spark with native library support
Date Thu, 06 Mar 2014 18:55:16 GMT
For the native libraries, you can use an existing Hadoop build and just put them on the path.
For linking to Hadoop, Spark grabs it through Maven, but you can do "mvn install" locally
on your version of Hadoop to install it to your local Maven cache, and then configure Spark
to use that version. Spark never builds Hadoop itself, it just downloads it through Maven.

Matei

On Mar 6, 2014, at 10:38 AM, Alan Burlison <Alan.Burlison@oracle.com> wrote:

> On 06/03/2014 17:44, Matei Zaharia wrote:
> 
>> Is it an error, or just a warning? In any case, you need to get those
>> libraries from a build of Hadoop for your platform. Then add them to
>> the SPARK_LIBRARY_PATH environment variable in conf/spark-env.sh, or
>> to your -Djava.library.path if launching an application separately.
> 
> OK, thanks. Is it possible to get Spark to build using an existing Hadoop build tree,
or does Spark insist on building its own Hadoop? The instructions at https://spark.incubator.apache.org/docs/latest/
seem to suggest that it always builds its own Hadoop version.
> 
> I may also have to fiddle with Hadoop to get it to build on Solaris if the instructions
at http://www.oracle.com/technetwork/articles/servers-storage-admin/sol-howto-native-hadoop-s11-1946524.html
are still relevant.
> 
>> These libraries just speed up some compression codecs BTW, so it
>> should be fine to run without them too.
> 
> Yes, it works as-is but I have a need for speed :-)
> 
> Thanks,
> 
> -- 
> Alan Burlison
> --


Mime
View raw message