Apache Spark 工作执行程序 EXITED，退出状态为 1

Question

我有一个包含 3 个工作人员的 Spark 独立设置 (v 1.4.1)。

我有一个应用程序，它从 Kafka 主题详细数据中读取流并将其存储在另一个 Kafka 主题中。

昨晚应用程序崩溃了，所有工作人员都崩溃了。

工人的日志报告如下：

16/02/04 21:02:10 INFO ExecutorRunner: Launch command: "/opt/jdk1.8.0_45/bin/java" "-cp" "/dati/spark-1.4.1-bin-hadoop2.4/sbin/../conf/:/dati/spark-1.4.1-bin-hadoop2.4/lib/spark-assembly-1.4.1-hadoop2.4.0.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-core-3.2.10.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-api-jdo-3.2.6.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-rdbms-3.2.9.jar" "-Xms1024M" "-Xmx1024M" "-Dspark.driver.port=52180" "-DenabledWorkerLog=false" "-Dcom.sun.management.jmxremote.port=54330" "-Dcom.sun.management.jmxremote.ssl=false" "-Dcom.sun.management.jmxremote.authenticate=false" "org.apache.spark.executor.CoarseGrainedExecutorBackend" "--driver-url" "akka.tcp://sparkDriver@driverHost:52180/user/CoarseGrainedScheduler" "--executor-id" "24279" "--hostname" "worker2" "--cores" "1" "--app-id" "app-20160201182749-0007" "--worker-url" "akka.tcp://sparkWorker@worker2:57853/user/Worker"
16/02/04 21:02:10 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160201182749-0007/24279/stdout with daily rolling
16/02/04 21:02:10 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160201182749-0007/24279/stderr with daily rolling
16/02/04 21:02:10 INFO Worker: Executor app-20160129184621-0001/1430 finished with state EXITED message Command exited with code 1 exitStatus 1
16/02/04 21:02:10 INFO Worker: Asked to launch executor app-20160129184621-0001/1431 for stream-elaboration
16/02/04 21:02:10 INFO ExecutorRunner: Launch command: "/opt/jdk1.8.0_45/bin/java" "-cp" "/dati/spark-1.4.1-bin-hadoop2.4/sbin/../conf/:/dati/spark-1.4.1-bin-hadoop2.4/lib/spark-assembly-1.4.1-hadoop2.4.0.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-core-3.2.10.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-api-jdo-3.2.6.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-rdbms-3.2.9.jar" "-Xms1024M" "-Xmx1024M" "-Dspark.driver.port=57297" "-DenabledWorkerLog=false" "-Dcom.sun.management.jmxremote.port=54326" "-Dcom.sun.management.jmxremote.ssl=false" "-Dcom.sun.management.jmxremote.authenticate=false" "org.apache.spark.executor.CoarseGrainedExecutorBackend" "--driver-url" "akka.tcp://sparkDriver@driverHost:57297/user/CoarseGrainedScheduler" "--executor-id" "1431" "--hostname" "worker2" "--cores" "1" "--app-id" "app-20160129184621-0001" "--worker-url" "akka.tcp://sparkWorker@worker2:57853/user/Worker"
16/02/04 21:02:10 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160129184621-0001/1431/stdout with daily rolling
16/02/04 21:02:10 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160129184621-0001/1431/stderr with daily rolling
16/02/04 21:02:11 INFO Worker: Executor app-20160201182749-0007/24279 finished with state EXITED message Command exited with code 1 exitStatus 1
16/02/04 21:02:11 INFO Worker: Asked to launch executor app-20160201182749-0007/24280 for stream-elaboration
16/02/04 21:02:11 INFO ExecutorRunner: Launch command: "/opt/jdk1.8.0_45/bin/java" "-cp" "/dati/spark-1.4.1-bin-hadoop2.4/sbin/../conf/:/dati/spark-1.4.1-bin-hadoop2.4/lib/spark-assembly-1.4.1-hadoop2.4.0.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-core-3.2.10.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-api-jdo-3.2.6.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-rdbms-3.2.9.jar" "-Xms1024M" "-Xmx1024M" "-Dspark.driver.port=52180" "-DenabledWorkerLog=false" "-Dcom.sun.management.jmxremote.port=54330" "-Dcom.sun.management.jmxremote.ssl=false" "-Dcom.sun.management.jmxremote.authenticate=false" "org.apache.spark.executor.CoarseGrainedExecutorBackend" "--driver-url" "akka.tcp://sparkDriver@driverHost:52180/user/CoarseGrainedScheduler" "--executor-id" "24280" "--hostname" "worker2" "--cores" "1" "--app-id" "app-20160201182749-0007" "--worker-url" "akka.tcp://sparkWorker@worker2:57853/user/Worker"
16/02/04 21:02:11 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160201182749-0007/24280/stdout with daily rolling
16/02/04 21:02:11 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160201182749-0007/24280/stderr with daily rolling
16/02/04 21:02:11 INFO Worker: Executor app-20160129184621-0001/1431 finished with state EXITED message Command exited with code 1 exitStatus 1
16/02/04 21:02:11 INFO Worker: Asked to launch executor app-20160129184621-0001/1432 for stream-elaboration
16/02/04 21:02:11 INFO ExecutorRunner: Launch command: "/opt/jdk1.8.0_45/bin/java" "-cp" "/dati/spark-1.4.1-bin-hadoop2.4/sbin/../conf/:/dati/spark-1.4.1-bin-hadoop2.4/lib/spark-assembly-1.4.1-hadoop2.4.0.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-core-3.2.10.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-api-jdo-3.2.6.jar:/dati/spark-1.4.1-bin-hadoop2.4/lib/datanucleus-rdbms-3.2.9.jar" "-Xms1024M" "-Xmx1024M" "-Dspark.driver.port=57297" "-DenabledWorkerLog=false" "-Dcom.sun.management.jmxremote.port=54326" "-Dcom.sun.management.jmxremote.ssl=false" "-Dcom.sun.management.jmxremote.authenticate=false" "org.apache.spark.executor.CoarseGrainedExecutorBackend" "--driver-url" "akka.tcp://sparkDriver@driverHost:57297/user/CoarseGrainedScheduler" "--executor-id" "1432" "--hostname" "worker2" "--cores" "1" "--app-id" "app-20160129184621-0001" "--worker-url" "akka.tcp://sparkWorker@worker2:57853/user/Worker"
16/02/04 21:02:11 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160129184621-0001/1432/stdout with daily rolling
16/02/04 21:02:11 INFO FileAppender: Rolling executor logs enabled for /dati/spark-1.4.1-bin-hadoop2.4/work/app-20160129184621-0001/1432/stderr with daily rolling
16/02/04 21:02:11 INFO Worker: Executor app-20160201182749-0007/24280 finished with state EXITED message Command exited with code 1 exitStatus 1
16/02/04 21:02:11 INFO Worker: Asked to launch executor app-20160201182749-0007/24281 for stream-elaboration

在日志的末尾：

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "qtp291507283-42"
Exception in thread "qtp291507283-37" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "ExecutorRunner for app-20160201182749-0007/29488" java.lang.OutOfMemoryError: GC overhead limit exceeded

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "sparkWorker-scheduler-1"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "qtp291507283-38" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "JMX server connection timeout 81" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "JMX server connection timeout 81"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "sparkWorker-10"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "qtp291507283-40"
Exception in thread "qtp291507283-35" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"

Exception in thread "qtp291507283-39" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "qtp291507283-41" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "qtp291507283-36" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" 
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "RMI TCP Connection(idle)"
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" java.lang.OutOfMemoryError: GC overhead limit exceeded
Exception in thread "RMI TCP Connection(idle)" 

....

跑步：

ps aux | grep "worker"

该进程仍然处于活动状态，但我在 sparkUI 上看不到它。

为什么worker executor重启的如此频繁？

Answer 1

日志显示多条java.lang.OutOfMemoryError: GC overhead limit exceeded 消息，这意味着您的执行程序会抛出导致它们退出的错误。

这个错误意味着你的程序在运行 GC 上花费了太多时间（更多细节见here）。要解决此问题 - 您可以尝试以下路径之一：

• 暴力破解方法是通过将 -XX:-UseGCOverheadLimit 添加到执行程序的 JVM 选项来禁用此安全性，但它可能会使您的应用程序主要执行 GC，因此运行速度非常慢
• 分析您的作业的内存使用情况并对其进行优化 - 您的代码可能会消耗比需要更多的内存，从而迫使 GC 过于努力地工作
• 调整内存设置 - 例如，如果您可以增加执行程序的堆空间，则可能会减少 GC 压力

Answer 2

您实际上是内存不足，无法顺利运行该过程。想到的选项：

1. 像您提到的那样指定更多内存，先尝试介于两者之间的东西，例如 -Xmx512m。
2. 调试代码以找出内存泄漏的可能性。

为什么worker executor重启的如此频繁？

您使用的是基于 Spark 构建的 Spark Streaming，因此它对工作节点具有相同的容错能力，这意味着如果工作节点由于某些意外错误而停机，如您的情况，Spark 引擎将尝试重新启动它。