spark-deep-learning
spark-deep-learning copied to clipboard
ERROR When running python unit tests
I run the unit tests command sparkdl$ SPARK_HOME=/usr/local/lib/spark-2.1.1-bin-hadoop2.7 PYSPARK_PYTHON=python2 SCALA_VERSION=2.11.8 SPARK_VERSION=2.1.1 ./python/run-tests.sh but I obtained the next error : py4j.java_gateway: DEBUG: Answer received: !xspy4j.Py4JException: Method listToVectorFunction([class org.apache.spark.sql.Column]) does not exist\n at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:318)\n at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:326)\n at py4j.Gateway.invoke(Gateway.java:272)\n at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)\n at py4j.commands.CallCommand.execute(CallCommand.java:79)\n at py4j.GatewayConnection.run(GatewayConnection.java:214)\n at java.lang.Thread.run(Thread.java:748)\n
How can I deal with this ?
It should be fixed in master now (with https://github.com/databricks/spark-deep-learning/pull/20) !
Please do reopen if you're still seeing problems. Thanks!
I am not using python3. I use python 2.7
@zsellami did you make some modifications to the scala part of the code? From the stack trace, a java object could be created but a method could not be found. You should make sure that the java part is recompiled by running build/sbt assembly
.
If this does not work, can you please upload the full stack trace using https://gist.github.com ?
I don't make any modification in the code. run build/sbt assembly work without any problem sparkdl$ SPARK_HOME=/usr/local/lib/spark-2.1.1-bin-hadoop2.7 PYSPARK_PYTHON=python2 SCALA_VERSION=2.11.8 SPARK_VERSION=2.1.1 ./python/run-tests.sh failed after some step. I upload the full stack trace of the shell. errorSparkDL.txt
@zsellami thank you for keeping updated with this issue.
Would you like to pull the master branch and try the unit test again?
git rev-parse HEAD
should return dee8fbff74c807bdbf72edf3588d4d5f64f36b0b
Please also run ./build/sbt assembly
and pip2 install -r python/requirements.txt
from the project root directory.
I tested with these locally on a clean branch and the all the tests passed.