Trying to run Spark Zeppelin
on Yarn
cluster, I set the following variables in zeppeline-env.sh
:
export HADOOP_CONF_DIR=/opt/hadoop/etc/hadoop
export SPARK_HOME=/opt/spark
export SPARK_SUBMIT_OPTIONS="--executor-memory 512m --driver-memory 512m --executor-cores 2 --queue onlineQ"
But it didn't use the SPARK_SUBMIT_OPTIONS
values in deploying job on YARN! For example executor-memory
is 1024m
as default. When I set executor-memory
in the Zeppelin interpreter
setting, it will be applied. As I should set --queue
property for Spark Submit, is there any other way (than SPARK_SUBMIT_OPTIONS variable) to set it?
You can set in spark interpreter settings.
spark.driver.memory
etc.
More information here: https://zeppelin.apache.org/docs/0.7.2/interpreter/spark.html#configuration