[SPARK-19814] Spark History Server Out Of Memory / Extreme GC

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 1.6.1, 2.0.0, 2.1.0
• Component/s: Web UI

Environment

Spark History Server (we've run it on several different Hadoop distributions)

• Assignee: Unassigned
• Reporter: Simon King
• Created: 03/Mar/17 19:53
• Updated: 08/Mar/17 22:17
• Resolved: 08/Mar/17 22:17

Description

Spark History Server runs out of memory, gets into GC thrash and eventually becomes unresponsive. This seems to happen more quickly with heavy use of the REST API. We've seen this with several versions of Spark.

Running with the following settings (spark 2.1):
spark.history.fs.cleaner.enabled true
spark.history.fs.cleaner.interval 1d
spark.history.fs.cleaner.maxAge 7d
spark.history.retainedApplications 500

We will eventually get errors like:
17/02/25 05:02:19 WARN ServletHandler:·
javax.servlet.ServletException: scala.MatchError: java.lang.OutOfMemoryError: GC overhead limit exceeded (of class java.lang.OutOfMemoryError)
at org.glassfish.jersey.servlet.WebComponent.serviceImpl(WebComponent.java:489)
at org.glassfish.jersey.servlet.WebComponent.service(WebComponent.java:427)
at org.glassfish.jersey.servlet.ServletContainer.service(ServletContainer.java:388)
at org.glassfish.jersey.servlet.ServletContainer.service(ServletContainer.java:341)
at org.glassfish.jersey.servlet.ServletContainer.service(ServletContainer.java:228)
at org.spark_project.jetty.servlet.ServletHolder.handle(ServletHolder.java:812)
at org.spark_project.jetty.servlet.ServletHandler.doHandle(ServletHandler.java:587)
at org.spark_project.jetty.server.handler.ContextHandler.doHandle(ContextHandler.java:1127)
at org.spark_project.jetty.servlet.ServletHandler.doScope(ServletHandler.java:515)
at org.spark_project.jetty.server.handler.ContextHandler.doScope(ContextHandler.java:1061)
at org.spark_project.jetty.server.handler.ScopedHandler.handle(ScopedHandler.java:141)
at org.spark_project.jetty.servlets.gzip.GzipHandler.handle(GzipHandler.java:529)
at org.spark_project.jetty.server.handler.ContextHandlerCollection.handle(ContextHandlerCollection.java:215)
at org.spark_project.jetty.server.handler.HandlerWrapper.handle(HandlerWrapper.java:97)
at org.spark_project.jetty.server.Server.handle(Server.java:499)
at org.spark_project.jetty.server.HttpChannel.handle(HttpChannel.java:311)
at org.spark_project.jetty.server.HttpConnection.onFillable(HttpConnection.java:257)
at org.spark_project.jetty.io.AbstractConnection$2.run(AbstractConnection.java:544)
at org.spark_project.jetty.util.thread.QueuedThreadPool.runJob(QueuedThreadPool.java:635)
at org.spark_project.jetty.util.thread.QueuedThreadPool$3.run(QueuedThreadPool.java:555)
at java.lang.Thread.run(Thread.java:745)

Caused by: scala.MatchError: java.lang.OutOfMemoryError: GC overhead limit exceeded (of class java.lang.OutOfMemoryError)
at org.apache.spark.deploy.history.ApplicationCache.getSparkUI(ApplicationCache.scala:148)
at org.apache.spark.deploy.history.HistoryServer.getSparkUI(HistoryServer.scala:110)
at org.apache.spark.status.api.v1.UIRoot$class.withSparkUI(ApiRootResource.scala:244)
at org.apache.spark.deploy.history.HistoryServer.withSparkUI(HistoryServer.scala:49)
at org.apache.spark.status.api.v1.ApiRootResource.getJobs(ApiRootResource.scala:66)
at sun.reflect.GeneratedMethodAccessor102.invoke(Unknown Source)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.glassfish.jersey.server.internal.routing.SubResourceLocatorRouter$1.run(SubResourceLocatorRouter.java:158)
at org.glassfish.jersey.server.internal.routing.SubResourceLocatorRouter.getResource(SubResourceLocatorRouter.java:178)
at org.glassfish.jersey.server.internal.routing.SubResourceLocatorRouter.apply(SubResourceLocatorRouter.java:109)
at org.glassfish.jersey.server.internal.routing.RoutingStage._apply(RoutingStage.java:109)
at org.glassfish.jersey.server.internal.routing.RoutingStage._apply(RoutingStage.java:112)
at org.glassfish.jersey.server.internal.routing.RoutingStage._apply(RoutingStage.java:112)
at org.glassfish.jersey.server.internal.routing.RoutingStage._apply(RoutingStage.java:112)
at org.glassfish.jersey.server.internal.routing.RoutingStage._apply(RoutingStage.java:112)
at org.glassfish.jersey.server.internal.routing.RoutingStage.apply(RoutingStage.java:92)
at org.glassfish.jersey.server.internal.routing.RoutingStage.apply(RoutingStage.java:61)
at org.glassfish.jersey.process.internal.Stages.process(Stages.java:197)
at org.glassfish.jersey.server.ServerRuntime$2.run(ServerRuntime.java:318)
at org.glassfish.jersey.internal.Errors$1.call(Errors.java:271)
at org.glassfish.jersey.internal.Errors$1.call(Errors.java:267)
at org.glassfish.jersey.internal.Errors.process(Errors.java:315)
at org.glassfish.jersey.internal.Errors.process(Errors.java:297)
at org.glassfish.jersey.internal.Errors.process(Errors.java:267)

at org.glassfish.jersey.process.internal.RequestScope.runInScope(RequestScope.java:317)
at org.glassfish.jersey.server.ServerRuntime.process(ServerRuntime.java:305)
at org.glassfish.jersey.server.ApplicationHandler.handle(ApplicationHandler.java:1154)
at org.glassfish.jersey.servlet.WebComponent.serviceImpl(WebComponent.java:473)

In our case we see memory usage gradually increase over perhaps 2 days, then level off near max heap size (4G in our case), then often within 12-24 hours GC activity will start to increase, and will result in more and more frequent errors, as in the stack trace above.

Issue Links

• relates to: SPARK-18085 SPIP: Better History Server scalability for many / large applications