Trystan Leftwich created SPARK-16017:
----------------------------------------
Summary: YarnClientSchedulerBackend now registers backends as IPs
instead of Hostnames which causes all tasks to run with RACK_LOCAL locality.
Key: SPARK-16017
URL: https://issues.apache.org/jira/browse/SPARK-16017
Project: Spark
Issue Type: Bug
Reporter: Trystan Leftwich
Priority: Critical
Since this change:
[SPARK-15395|https://issues.apache.org/jira/browse/SPARK-15395]
When registering new executor backends it registers them as IPs instead of
hostnames. This causes a flow on effect that when the Task manager is trying to
figure out what Locality tasks should run at, no tasks can be run At the
NODE_LOCAL level.
This specific call:
https://github.com/apache/spark/blob/branch-2.0/core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala#L886
[pendingTasksForHost|https://github.com/apache/spark/blob/branch-2.0/core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala#L886]
are all hostnames pulled from the DFS locations while
[hasExecutorsAliveOnHost|https://github.com/apache/spark/blob/9b234b55d1b5e4a7c80e482b3e297bfb8b583a56/core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala#L560],
uses
[executorsByHost|https://github.com/apache/spark/blob/branch-2.0/core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala#L98],
which are all IP's because they are populated from the RpcAddress.
As expected this causes significant performance problems, A simple count query
will take 22 seconds, But if I revert the change from
[SPARK-15395|https://issues.apache.org/jira/browse/SPARK-15395], tasks will run
with NODE_LOCAL locality and the same count will take 3 seconds.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
