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 >>>>> >>>>