Badri Krishnan created SPARK-27511:
--------------------------------------
Summary: Spark Streaming Driver Memory
Key: SPARK-27511
URL: https://issues.apache.org/jira/browse/SPARK-27511
Project: Spark
Issue Type: Question
Components: DStreams
Affects Versions: 2.4.0
Reporter: Badri Krishnan
Hello Apache Spark Community.
We are currently facing an issue with one of our Spark Streaming jobs which
consumes data from a IBM MQ, this is run on a AWS EMR cluster using DStreams
and Checkpointing.
Our Spark streaming job failed with several containers exiting with error code:
143. I checked your container logs. For example, one of the killed container's
stdout logs [1] show the below error: (Exit code from container
container_1553356041292_0001_15_000004 is : 143)
2019-03-28 19:32:26,569 ERROR [dispatcher-event-loop-3]
org.apache.spark.streaming.receiver.ReceiverSupervisorImpl:Error stopping
receiver 2 org.apache.spark.SparkException: Exception thrown in awaitResult:
at org.apache.spark.util.ThreadUtils$.awaitResult(ThreadUtils.scala:226)
....
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.io.IOException: Failed to connect to
ip-**-***-*.***.***.com/**.**.***.**:*****
at
org.apache.spark.network.client.TransportClientFactory.createClient(TransportClientFactory.java:245)
at
org.apache.spark.network.client.TransportClientFactory.createClient(TransportClientFactory.java:187)
at org.apache.spark.rpc.netty.NettyRpcEnv.createClient(NettyRpcEnv.scala:198)
at org.apache.spark.rpc.netty.Outbox$$anon$1.call(Outbox.scala:194)
at org.apache.spark.rpc.netty.Outbox$$anon$1.call(Outbox.scala:190)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
... 3 more
These containers exited with code 143 because it was not able to reach the
application master(Driver Process).
Amazon mentioned that the Application Master is consuming more memory and hence
recommended us to double it. As AM runs on driver, we were asked to increase
spark.driver.memory from 1.4G to 3G. But the question that was unanswered was
whether increasing the memory would solve the problem or delay the failure. As
this is an ever running streaming application, do we need to consider something
to understand whether the memory usage builds up over a period of time or are
there any properties that needs to be set specific to how AM(application
Master) works for streaming application. Any inputs on how to track the AM
memory usage? Any insights will be helpful.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
