You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@beam.apache.org by Shrijit Pillai <pi...@gmail.com> on 2019/01/03 19:50:44 UTC
Child jobs not kicking off using SparkRunner in cluster mode
Hello,
I'm trying to run the WordCount example using SparkRunner.
In the client-mode, the child jobs are kicked off and the output is also produced. However in the cluster mode, the child jobs are not starting and no output is produced.
I'm using Beam 2.9.0 and Spark 2.3.0
Here are the commands that I ran:-
Cluster-mode (No output produced)
spark2-submit --conf spark.driver.userClassPathFirst=true --class org.wordcount.WordCountSpark --master yarn --queue test --deploy-mode cluster wordcount-0.1.0-SNAPSHOT-shaded.jar --runner=SparkRunner
Client-mode (Output produced)
spark2-submit --conf spark.driver.userClassPathFirst=true --class org.wordcount.WordCountSpark --master yarn --queue test --deploy-mode client wordcount-0.1.0-SNAPSHOT-shaded.jar --runner=SparkRunner
Here's the code
https://gist.github.com/shrijitpillai/852a3c737c922af9e9ba4cd4c53b8530
Here's the mvn dependency tree:-
https://gist.github.com/shrijitpillai/f06797238cf6db9bf350e6a835fb34d6
Cluster mode yarn log
https://gist.github.com/shrijitpillai/2f1963ece26c152a2d68bc6ae8913fea
Can somebody help in identifying what is that I'm missing here?
Thanks,
Shrijit Pillai