[LIVY-602] An error during 'magic' (table / json) encoding of a result

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.6.0
• Fix Version/s: 0.9.0
• Component/s: Interpreter
• Assignee: Unassigned
• Reporter: Maksim Ochenashko
• Created: 27/Jun/19 16:45
• Updated: 12/Nov/22 22:09

Description

The error occurs during the execution of the statement:

val x = List((1, "a"), (3, "b")).toDF("number", "key")
%table x //or %json x

Environment:
Docker
Livy: 0.6-incubating
Spark: 2.4.0
Hadoop: 3.0
Dockerfile:

FROM gettyimages/spark:2.4.0-hadoop-3.0

RUN apt-get update && \
    apt-get install -y wget && \
    rm -rf /var/lib/apt/lists/*

RUN mkdir -p /apps && \
    cd /apps && \
    wget http://apache.cp.if.ua/incubator/livy/0.6.0-incubating/apache-livy-0.6.0-incubating-bin.zip && \
    unzip apache-livy-0.6.0-incubating-bin.zip -d /apps && \
    rm -f apache-livy-0.6.0-incubating-bin.zip && \
    mkdir /apps/apache-livy-0.6.0-incubating-bin/logs

# Expose port
EXPOSE 8998

CMD ["/apps/apache-livy-0.6.0-incubating-bin/bin/livy-server"]

Reproducing steps (bash):

docker build . -t livy-spark

docker run -d -p 8998:8998 livy-spark

curl localhost:8998/sessions -X POST --data '{"kind":"spark"}' -H "Content-Type: application/json"

curl localhost:8998/sessions/0/statements -X POST -H 'Content-Type: application/json' -d '{"code":"val x = List((1, \"a\"), (3, \"b\")).toDF(\"number\", \"key\") \r\n%table x"}'

curl localhost:8998/sessions/0/statements/0

// response
"id":0,"code":"val x = List((1, \"a\"), (3, \"b\")).toDF(\"number\", \"key\") \n%table x","state":"available","output":{"status":"error","execution_count":0,"ename":"Internal Error: java.lang.NoClassDefFoundError","evalue":"scala/tools/scalap/scalax/rules/scalasig/Type","traceback":[]},"progress":1.0}%

curl localhost:8998/sessions/0/statements -X POST -H 'Content-Type: application/json' -d '{"code":"val x = List((1, \"a\"), (3, \"b\")).toDF(\"number\", \"key\") \r\n%json x"}'

curl localhost:8998/sessions/0/statements/1

// response
{"id":1,"code":"val x = List((1, \"a\"), (3, \"b\")).toDF(\"number\", \"key\") \n%json x","state":"available","output":{"status":"error","execution_count":1,"ename":"ValueError","evalue":"Failed to convert value into a JSON value","traceback":[]},"progress":1.0}%

The error:
Livy cannot load the scala/tools/scalap/scalax/rules/scalasig/Type class.

The line causing the error:
https://github.com/apache/incubator-livy/blob/v0.6.0-incubating/repl/src/main/scala/org/apache/livy/repl/AbstractSparkInterpreter.scala#L186

Stacktrace: 
https://gist.github.com/iRevive/977a67940b05ab2faaa3f3be5b426038

Possible root cause:
Looks like there is a problem with the classpath.
Currently, Livy has scalap-2.11.0.jar in the classpath. I tried to replace it with scalap-2.11.12.jar but it still doesn't work.