Hi Thomas, I try to the latest branch-0.8, it works for me. Could you try again to verify it ?
Thomas Bünger <thom.bu...@googlemail.com>于2018年6月7日周四 下午8:34写道: > I specifically mean visualisation via ZeppelinContext inside a Spark > interpreter. (e.g. "z.show(...)") > The visualisation of SparkSQL results inside a SparkSQLInterpreter work > fine, also in yarn-cluster mode. > > Am Do., 7. Juni 2018 um 14:30 Uhr schrieb Thomas Bünger < > thom.bu...@googlemail.com>: > >> Hey Jeff, >> >> I tried your changes and now it works nicely. Thank you very much! >> >> But I still can't use any of the forms and visualizations in yarn-cluster? >> I was hoping that this got resolved with the new SparkInterpreter so that >> I can switch from yarn-client to yarn-cluster mode in 0.8, but I'm still >> getting errors like >> "error: not found: value z" >> >> Was this not in scope of that change? Is this a bug? Or is it known >> limitation and also not supported in 0.8? >> >> Best regards, >> Thomas >> >> Am Mi., 6. Juni 2018 um 03:28 Uhr schrieb Jeff Zhang <zjf...@gmail.com>: >> >>> >>> I can confirm that this is a bug, and created >>> https://issues.apache.org/jira/browse/ZEPPELIN-3531 >>> >>> Will fix it soon >>> >>> Jeff Zhang <zjf...@gmail.com>于2018年6月5日周二 下午9:01写道: >>> >>>> >>>> hmm, it looks like a bug. I will check it tomorrow. >>>> >>>> >>>> Thomas Bünger <thom.bu...@googlemail.com>于2018年6月5日周二 下午8:56写道: >>>> >>>>> $ ls /usr/lib/spark/python/lib >>>>> py4j-0.10.6-src.zip PY4J_LICENSE.txt pyspark.zip >>>>> >>>>> So folder exists and contains both necessary zips. Please note, that >>>>> in local or yarn-client mode the files are properly picked up from that >>>>> very same location. >>>>> >>>>> How does yarn-cluster work under the hood? Could it be that >>>>> environment variables (like SPARK_HOME) are lost, because they are only >>>>> available in my local shell + zeppelin daemon process? Do I need to tell >>>>> YARN somehow about SPARK_HOME? >>>>> >>>>> Am Di., 5. Juni 2018 um 14:48 Uhr schrieb Jeff Zhang <zjf...@gmail.com >>>>> >: >>>>> >>>>>> >>>>>> Could you check whether there's folder /usr/lib/spark/python/lib ? >>>>>> >>>>>> >>>>>> Thomas Bünger <thom.bu...@googlemail.com>于2018年6月5日周二 下午8:45写道: >>>>>> >>>>>>> >>>>>>> sys.env >>>>>>> java.lang.NullPointerException at >>>>>>> org.apache.zeppelin.spark.NewSparkInterpreter.setupConfForPySpark(NewSparkInterpreter.java:149) >>>>>>> at >>>>>>> org.apache.zeppelin.spark.NewSparkInterpreter.open(NewSparkInterpreter.java:90) >>>>>>> at >>>>>>> org.apache.zeppelin.spark.SparkInterpreter.open(SparkInterpreter.java:62) >>>>>>> at >>>>>>> org.apache.zeppelin.interpreter.LazyOpenInterpreter.open(LazyOpenInterpreter.java:69) >>>>>>> at >>>>>>> org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer$InterpretJob.jobRun(RemoteInterpreterServer.java:617) >>>>>>> at org.apache.zeppelin.scheduler.Job.run(Job.java:188) at >>>>>>> org.apache.zeppelin.scheduler.FIFOScheduler$1.run(FIFOScheduler.java:140) >>>>>>> at >>>>>>> java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) >>>>>>> at java.util.concurrent.FutureTask.run(FutureTask.java:266) at >>>>>>> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180) >>>>>>> at >>>>>>> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) >>>>>>> at >>>>>>> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) >>>>>>> at >>>>>>> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) >>>>>>> at java.lang.Thread.run(Thread.java:748) >>>>>>> >>>>>>> >>>>>>> Am Di., 5. Juni 2018 um 14:41 Uhr schrieb Jeff Zhang < >>>>>>> zjf...@gmail.com>: >>>>>>> >>>>>>>> Could you paste the full stracktrace ? >>>>>>>> >>>>>>>> >>>>>>>> Thomas Bünger <thom.bu...@googlemail.com>于2018年6月5日周二 下午8:21写道: >>>>>>>> >>>>>>>>> I've tried the 0.8.0-rc4 on my EMR cluster using the preinstalled >>>>>>>>> version of spark under /usr/lib/spark. >>>>>>>>> >>>>>>>>> This works fine in local or yarn-client mode, but in yarn-cluster >>>>>>>>> mode i just get a >>>>>>>>> >>>>>>>>> java.lang.NullPointerException at >>>>>>>>> org.apache.zeppelin.spark.NewSparkInterpreter.setupConfForPySpark(NewSparkInterpreter.java:149) >>>>>>>>> >>>>>>>>> Seems to be caused by an unsuccessful search for the py4j >>>>>>>>> libraries. >>>>>>>>> I've made sure that SPARK_HOME is actually set in .bash_rc, in >>>>>>>>> zeppelin-env.sh and via the new %spark.conf, but somehow in the remote >>>>>>>>> interpreter, something odd is going on. >>>>>>>>> >>>>>>>>> Best regards, >>>>>>>>> Thomas >>>>>>>>> >>>>>>>>