spark.yarn.executor.memoryOverhead

[#1]

Got below error

17/09/12 20:41:36 WARN yarn.YarnAllocator: Container killed by YARN for exceeding memory limits. 1.5 GB of 1.5 GB physical memory used. Consider boosting spark.yarn.executor.memoryOverhead.
17/09/12 20:41:39 ERROR cluster.YarnClusterScheduler: Lost executor 1 on xyz.com: remote Akka client disassociated

Please help as not able to find spark.executor.memory or spark.yarn.executor.memoryOverhead in Cloudera Manager (Cloudera Enterprise 5.4.7)

[#2]

These can be set globally, try searching for just spark memory as CM doesn't always include the actual setting name.

These can be set per job as well. Spark-submit --executor-memory

https://spark.apache.org/docs/1.6.0/submitting-applications.html

[#3 · Accepted solution]

spark.executor.memory  can be found in Cloudera Manager under Hive->configuration and search for Java Heap.

Spark Executor Maximum Java Heap Size
spark.executor.memory
HiveServer2 Default Group

256 MiB

Spark Driver Maximum Java Heap Size
spark.driver.memory
HiveServer2 Default Group

256 MiB

[#4]

Thank you.
Additional query, do you know why these spark configs are placed under hive?

[#5]

It's a spark side configuraion. So you can always specify it via "--conf" option with spark-submit, or you can set the property globally on CM via "Spark Client Advanced Configuration Snippet (Safety Valve) for spark-conf/spark-defaults.conf", so CM will include such setting for you via spark gateway client configuration.

[#6]

Hi,

This parameter spark.executor.memory (or) spark.yarn.executor.memoryOverhead can be set in Spark submit command or you can set it Advanced configurations.

Thanks

AKR