Hi David, With the master branch (0.6.0-SNAPSHOT), 1) download and configure spark and make sure everything works with bin/spark-shell command. If you have extra jar, you can add spark.files property in SPARK_HOME/conf/spark-defaults.conf 2) export SPARK_HOME in conf/zeppelin-env.sh file. 3) Start Zeppelin and configure master, spark.executor.memory, etc in Interpreter menu.
I think these steps are easiest way to set up Zeppelin with extra jars. Thanks, moon On Thu, Sep 17, 2015 at 5:41 PM David Salinas <david.salinas....@gmail.com> wrote: > Hi, > > I have just tested without setting the classpath in the conf and 1/ > worked. Could you tell me what is the best way to set up your classpath/jar > now? > > Best, > > David > > On Thu, Sep 17, 2015 at 10:34 AM, David Salinas < > david.salinas....@gmail.com> wrote: > >> Hi, >> >> I have tried this example after >> https://github.com/apache/incubator-zeppelin/pull/270. >> >> But it is not working for several reasons: >> >> 1/ Zeppelin context is not found (!): >> val s = z.input("Foo") >> <console>:21: error: not found: value z >> >> 2/ If I include my jar, the classpath is not communicated to slaves, the >> code works only locally (it used to work on the cluster before this >> change), I guess there is something wrong with the way I set the classpath >> (which is also probably linked to 1/) >> >> I have added this line in zeppelin-env.sh to use one of my jar >> export ZEPPELIN_JAVA_OPTS="-Dspark.driver.host=`hostname` >> -Dspark.mesos.coarse=true -Dspark.executor.memory=20g -Dspark.cores.max=80 >> -Dspark.jars=${SOME_JAR} -cp ${SOME_CLASSPATH_FOR_THE_JAR} >> >> How can one add its extraclasspath jar with this new version? Could you >> add an ZEPPELIN_EXTRA_JAR, ZEPPELIN_EXTRA_CLASSPATH to zeppelin-env.sh so >> that the user can add easily his code? >> >> Best, >> >> David >> > >
