Zeppelin currently propagates a jar including `SparkInterprer`. Thus if you add some dependencies into interpreter tab of Spark, that jar doesn't pass to executors so that error may occurs. How about using --packages option?
On Mon, Oct 24, 2016 at 10:32 PM, herman...@teeupdata.com < herman...@teeupdata.com> wrote: > yes, yarn-client model. > > In situation of spark streaming, does zeppelin support graphs that are > automatically update themselves, like a self-updated dashboard on top of > streaming based spark tables? > > Thanks > Herman. > > > On Oct 21, 2016, at 23:42, Jongyoul Lee <jongy...@gmail.com> wrote: > > Hi, > > Do you use yarn-client mode of Spark? > > On Friday, 21 October 2016, herman...@teeupdata.com < > herman...@teeupdata.com> wrote: > >> Hi Everyone, >> >> Does zeppelin support spark streaming with kafka? I am using zeppelin >> 0.6.1 with spark 2.0 and kafka 0.10.0.0. >> >> I got error when import org.apache.spark.streaming.kafka.KafkaUtils >> <console>:36: error: object kafka is not a member of package >> org.apache.spark.streaming >> import org.apache.spark.streaming.kafka.KafkaUtils >> >> I already added the spark streaming kafka jar to the dependencies of >> spark interpreter. >> >> If it is supported, is there a tutorial/sample notebook? >> >> Thanks >> Herman. >> >> >> >> > > -- > 이종열, Jongyoul Lee, 李宗烈 > http://madeng.net > > > -- 이종열, Jongyoul Lee, 李宗烈 http://madeng.net
