Jeff Zhang created SPARK-17125:
----------------------------------
Summary: Allow to specify spark config using non-string type in
SparkR
Key: SPARK-17125
URL: https://issues.apache.org/jira/browse/SPARK-17125
Project: Spark
Issue Type: Improvement
Components: SparkR
Affects Versions: 2.0.0
Reporter: Jeff Zhang
Priority: Minor
I try to specify spark conf spark.executor.instances as following in SparkR,
but fails. Since list supports any kind of data type, it is natural for user to
specify int type for configuration like spark.exeucotr.instances.
{code}
sparkR.session(master="yarn-client", sparkConfig =
list(spark.executor.instances=1))
{code}
{noformat}
Error in invokeJava(isStatic = TRUE, className, methodName, ...) :
java.lang.IllegalArgumentException: spark.executor.instances should be int,
but was 1.0
at
org.apache.spark.internal.config.ConfigHelpers$.toNumber(ConfigBuilder.scala:31)
at
org.apache.spark.internal.config.ConfigBuilder$$anonfun$intConf$1.apply(ConfigBuilder.scala:178)
at
org.apache.spark.internal.config.ConfigBuilder$$anonfun$intConf$1.apply(ConfigBuilder.scala:178)
at scala.Option.map(Option.scala:146)
at
org.apache.spark.internal.config.OptionalConfigEntry.readFrom(ConfigEntry.scala:150)
at
org.apache.spark.internal.config.OptionalConfigEntry.readFrom(ConfigEntry.scala:138)
at org.apache.spark.SparkConf.get(SparkConf.scala:251)
at
org.apache.spark.deploy.yarn.YarnSparkHadoopUtil$.getInitialTargetExecutorNumber(YarnSparkHadoopUtil.scala:313)
at
org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.start(YarnClientSchedulerBackend.scala:54)
at
org.apache.spark.scheduler.TaskSchedulerImpl.start(TaskSchedulerImpl.scala:154)
at org.apache.spark.SparkContext
{noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
