Hi,
I have met with a strange issue in running R notebooks in zeppelin(0.7.2).
Spark intrepreter is in per note Scoped mode and spark version is 1.6.2
Please find the steps below to reproduce the issue:
1. Create a notebook (Note1) and run any r code in a paragraph. I ran the
following code.
> %r
>
> rdf <- data.frame(c(1,2,3,4))
>
> colnames(rdf) <- c("myCol")
>
> sdf <- createDataFrame(sqlContext, rdf)
>
> withColumn(sdf, "newCol", sdf$myCol * 2.0)
>
>
2. Create another notebook (Note2) and run any r code in a paragraph. I
ran the same code as above.
Till now everything works fine.
3. Create third notebook (Note3) and run any r code in a paragraph. I ran
the same code. This notebook fails with the error
> org.apache.zeppelin.interpreter.InterpreterException: sparkr is not
> responding
What I understood from the analysis is that the process created for
sparkr interpreter is not getting killed properly and this makes every
third model to throw an error while executing. The process will be killed
on restarting the sparkr interpreter and another 2 models could be executed
successfully. ie, For every third model run using the sparkr interpreter,
the error is thrown. We suspect this as a limitation with zeppelin.
Please help to solve this issue
Regards,
Meethu Mathew
