[
https://issues.apache.org/jira/browse/SPARK-2614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14070908#comment-14070908
]
Mark Hamstra commented on SPARK-2614:
-------------------------------------
It's also common for installers/admins to not want all of the examples
installed on their production machines.
This issue is really touching on the fact that Spark's Debian packaging isn't
presently everything that it could be. It's really just a hack to allow
deployment tools like Chef to be able to manage Spark. To really do Debian
packaging right, we should be creating multiple packages -- perhaps spark-core,
spark-sql, spark-mllib, spark-graphx, spark-examples, etc. The only thing that
is really preventing us from doing this is that proper Debian packaging would
require a package maintainer willing and committed to do all of the maintenance
work.
> Add the spark-examples-xxx-.jar to the Debian package created by
> assembly/pom.xml (e.g. -Pdeb)
> ----------------------------------------------------------------------------------------------
>
> Key: SPARK-2614
> URL: https://issues.apache.org/jira/browse/SPARK-2614
> Project: Spark
> Issue Type: Improvement
> Components: Build, Deploy
> Reporter: Christian Tzolov
>
> The tar.gz distribution includes already the spark-examples.jar in the
> bundle. It is a common practice for installers to run SparkPi as a smoke test
> to verify that the installation is OK
> /usr/share/spark/bin/spark-submit \
> --num-executors 10 --master yarn-cluster \
> --class org.apache.spark.examples.SparkPi \
> /usr/share/spark/jars/spark-examples-1.0.1-hadoop2.2.0.jar 10
--
This message was sent by Atlassian JIRA
(v6.2#6252)
