Spark on Yarn job fails when launching container

Spark on Yarn job fails when launching container

Symptom:

When running Spark jobs on Yarn: 
bin/spark-submit --class org.apache.spark.examples.SparkPi --master yarn-cluster lib/spark-examples*.jar 10
the job fails with below error message from resource manager log:
INFO org.apache.hadoop.yarn.server.resourcemanager.rmapp.RMAppImpl: Application application_1417043690539_0013 failed 2 times due to AM Container for appattempt_1417043690539_0013_000002 exited with  exitCode: 1 due to: Exception from container-launch:
org.apache.hadoop.util.Shell$ExitCodeException:
        at org.apache.hadoop.util.Shell.runCommand(Shell.java:505)
        at org.apache.hadoop.util.Shell.run(Shell.java:418)
        at org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:650)
        at org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor.launchContainer(LinuxContainerExecutor.java:295)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:314)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:83)
        at java.util.concurrent.FutureTask.run(FutureTask.java:262)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        at java.lang.Thread.run(Thread.java:744)

main : command provided 1
main : user is root
main : requested yarn user is root

Container exited with a non-zero exit code 1
.Failing this attempt.. Failing the application.

Root Cause:

To find the root cause, we can follow below troubleshooting path, especially when Yarn log aggregation is not enabled.(By default, yarn.log-aggregation-enable=false)

1. Check which node manager has the failure from resource manager log.

By default, both resource manager log and node manager log are located at /opt/mapr/hadoop/hadoop-2.4.1/logs .
In this case, this attempt is on node "yarn-fcs-2":
INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler: 
Application attempt appattempt_1417043690539_0013_000002 released container 
container_1417043690539_0013_02_000001 on node: host: yarn-fcs-2:42846

2. Check which container has the failure and what is the error message in node manager log.

In this case, the container is container_1417043690539_0013_02_000001 and the error message in node manager log is :
WARN org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: 
Exception from container-launch with container ID: container_1417043690539_0013_02_000001 and exit code: 1
org.apache.hadoop.util.Shell$ExitCodeException:
        at org.apache.hadoop.util.Shell.runCommand(Shell.java:505)
        at org.apache.hadoop.util.Shell.run(Shell.java:418)
        at org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:650)
        at org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor.launchContainer(LinuxContainerExecutor.java:295)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:314)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:83)
        at java.util.concurrent.FutureTask.run(FutureTask.java:262)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        at java.lang.Thread.run(Thread.java:744)
From above error message, we know this issue happened when launching container on this node.
However above error message does not tell the reason.

3. Check container stdout and stderr for the reason of the failure.

The container log is determined by parameter "yarn.nodemanager.log-dirs" in yarn-site.xml.
By default, it is set to ${yarn.log.dir}/userlogs, which means $HADOOP_YARN_HOME/logs/userlogs.
In this case, the container log is located here:
/opt/mapr/hadoop/hadoop-2.4.1/logs/userlogs/application_1417043690539_0013/container_1417043690539_0013_02_000001
The reason of the failure is :
# cat stderr
Error: Could not find or load main class org.apache.spark.deploy.yarn.ExecutorLauncher

4. Find out which jar file has this missing class.

By searching, the missing jar file is located here:
/opt/mapr/spark/spark-1.1.0-bin-2.4.1-mapr-1408/lib/spark-assembly-1.1.0-hadoop2.4.1-mapr-1408.jar
This jar file is very important for running Spark on Yarn.

Solution:

1. Put spark-assembly-<version>.jar in property "yarn.application.classpath" from yarn-site.xml.
<property>
    <name>yarn.application.classpath</name>
    <value>/opt/mapr/spark/spark-1.1.0-bin-2.4.1-mapr-1408/lib/spark-assembly-1.1.0-hadoop2.4.1-mapr-1408.jar,/opt/mapr/hadoop/hadoop-2.4.1/etc/hadoop,/opt/mapr/hadoop/hadoop-2.4.1/etc/hadoop,/opt/mapr/hadoop/hadoop-2.4.1/etc/hadoop,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/common/lib/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/common/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/hdfs,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/hdfs/lib/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/hdfs/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/yarn/lib/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/yarn/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/mapreduce/lib/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/mapreduce/*,/contrib/capacity-scheduler/*.jar,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/yarn/*,/opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/yarn/lib/*
    </value>
</property>
2. Restart resource manager and node managers.
After that, the query works fine.

Commentaires

Enregistrer un commentaire

Posts les plus consultés de ce blog

Controlling Parallelism in Spark by controlling the input partitions by controlling the input partitions

Image
It is useful to be able to control the degree of parallelism when using Spark. Spark provides a very convenient method to increase the degree of parallelism which should be adequate in practice. This blog entry of the "Inside Spark" series describes the knobs Spark uses to control the degree of parallelism. Controlling the number of partitions in the " local " mode Spark is designed ground up to enable unit testing. To control the degree of parallelism in the local mode simply utilize the function JavaRDD<String> org.apache.spark.api.java.JavaSparkContext.parallelize(List<String> list, int numSlices) The following snippet of code demonstrates this usage- public class SparkControlPartitionSizeLocally { public static void main ( String [] args ) { JavaSparkContext sc = new JavaSparkContext ( "local" , "localpartitionsizecontrol" ); String input = "four score and sev
Lire la suite

Spark performance optimization: shuffle tuning

Image
Article directory 1 shuffle tuning 1.1 Summary of tuning 1.2 Overview of ShuffleManager Development 1.3 HashShuffleManager operating principle 1.3.1 Unoptured HashShuffleManager 1.3.2 Optimized HashShuffleManager 1.4 SortShuffleManager operating principle 1.4.1 General operating mechanism 1.4.2 bypass running mechanism 1.5 shuffle related parameters tuning 1.5.1 spark.shuffle.file.buffer 1.5.2 spark.reducer.maxSizeInFlight 1.5.3 spark.shuffle.io.maxRetries 1.5.4 spark.shuffle.io.retryWait 1.5.5 spark.shuffle.memoryFraction 1.5.6 spark.shuffle.manager 1.5.7 spark.shuffle.sort.bypassMergeThreshold 1.5.8 spark.shuffle.consolidateFiles 2 write in the last words Shuffle Summary of tuning Most of the performance of Spark operations is mainly consumed in the shuffle link, because the link contains a large number of disk IO, serialization, network data transmission and othe
Lire la suite

Spark optimization

Image
One operation and maintenance 1. Master hang up, standby restart is also invalid Master defaults to 512M of memory, when the task in the cluster is particularly high, it will hang, because the master will read each task event log log to generate spark ui, the memory will naturally OOM, you can run the log See that the master of the start through the HA will naturally fail for this reason. solve Increase the Master's memory spark-env.sh , set in the master node spark-env.sh : export SPARK_DAEMON_MEMORY 10g # 根据你的实际情况 Reduce the job information stored in the Master memory spark.ui.retainedJobs 500 # 默认都是1000 spark.ui.retainedStages 500 Hang up or suspend Sometimes we will see the web node in the web ui disappear or in the dead state, the task of running the node will report a variety of lost worker errors, causing the same reasons and the above, worker memory to save a lot of ui The information leads to gc when the heartbeat is lost
Lire la suite