Could you check the classpath of the interpreter process ? I suspect you
didn't export HADOOP_CONF_DIR correctly.
Run the following command:
ps aux | grep RemoteInterpreterServer
On Sun, Oct 9, 2016 at 2:36 PM, Xi Shen <davidshe...@gmail.com> wrote:
> Hi,
>
> I followed http://zeppelin.apache.org/docs/latest/interpreter/spark.html,
> and set up SPARK_HOME, HADOOP_CONF_DIR.
>
> My SPARK build is 2.0. My Zeppelin build is the 0.6.1 binary from the web.
>
> After I start Zeppelin, I went to the interpreter setting page and changed
> Spark interpreter settings as follow:
>
> master: yarn
> deploy-mode: client
>
> Then, I created a new notebook and executed:
>
> %spark
>
> spark.version
>
> The block never finishes...no error either.
>
> In the ./logs/zeppelin-interpreter-spark-*.log, I found the following *which
> I think is the cause of my problem*
>
> INFO [2016-10-09 06:28:37,074] ({pool-2-thread-2}
> Logging.scala[logInfo]:54) - Added JAR
> *file:/opt/zeppelin-0.6.1-bin-all/interpreter/spark/zeppelin-spark_2.11-0.6.1.jar*
> at spark://172.17.0.3:38775/jars/zeppelin-spark_2.11-0.6.1.jar with
> timestamp 1475994517073
> INFO [2016-10-09 06:28:37,150] ({pool-2-thread-2}
> Logging.scala[logInfo]:54) - Created default pool default, schedulingMode:
> FIFO, minShare: 0, weight: 1
> INFO [2016-10-09 06:28:38,205] ({pool-2-thread-2}
> RMProxy.java[createRMProxy]:98) - *Connecting to ResourceManager at
> /0.0.0.0:8032 <http://0.0.0.0:8032>*
>
> Looks like Zeppelin is not using my Spark binary, and is not using my
> hadoop configuration.
>
> What did I miss?
>
> --
>
>
> Thanks,
> David S.
>
--
Best Regards
Jeff Zhang
