From 435b6ae82b7c8b5fb3e8ea830a6972f03b3b8cf6 Mon Sep 17 00:00:00 2001 From: Jason Lowe Date: Mon, 25 Jan 2016 16:11:59 +0000 Subject: [PATCH] YARN-4452. NPE when submit Unmanaged application. Contributed by Naganarasimha G R. (cherry picked from commit 50bd067e1d63d4c80dc1e7bf4024bfaf42cf4416) Conflicts: hadoop-yarn-project/CHANGES.txt --- hadoop-yarn-project/CHANGES.txt | 9 +++++ .../metrics/SystemMetricsPublisher.java | 22 +++++++----- .../metrics/TestSystemMetricsPublisher.java | 34 +++++++++++++++---- 3 files changed, 51 insertions(+), 14 deletions(-) diff --git a/hadoop-yarn-project/CHANGES.txt b/hadoop-yarn-project/CHANGES.txt index bde7e671f6d..c828d32dea4 100644 --- a/hadoop-yarn-project/CHANGES.txt +++ b/hadoop-yarn-project/CHANGES.txt @@ -1100,6 +1100,9 @@ Release 2.8.0 - UNRELEASED YARN-4402. TestNodeManagerShutdown And TestNodeManagerResync fails with bind exception. (Brahma Reddy Battula via jianhe) + YARN-4452. NPE when submit Unmanaged application. (Naganarasimha G R + via junping_du) + YARN-4392. ApplicationCreatedEvent event time resets after RM restart/failover. (Naganarasimha G R and Xuan Gong via xgong) @@ -1211,6 +1214,9 @@ Release 2.7.3 - UNRELEASED YARN-4439. Clarify NMContainerStatus#toString method. (Jian He via xgong) + YARN-4452. NPE when submit Unmanaged application. (Naganarasimha G R via + junping_du) + YARN-4398. Remove unnecessary synchronization in RMStateStore. (Ning Ding via jianhe) YARN-4422. Generic AHS sometimes doesn't show started, node, or logs on App page @@ -2092,6 +2098,9 @@ Release 2.6.4 - UNRELEASED BUG FIXES + YARN-4452. NPE when submit Unmanaged application. (Naganarasimha G R + via junping_du) + YARN-4546. ResourceManager crash due to scheduling opportunity overflow. (Jason Lowe via junping_du) diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/SystemMetricsPublisher.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/SystemMetricsPublisher.java index 2c6349cb98e..84a3b1992d6 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/SystemMetricsPublisher.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/SystemMetricsPublisher.java @@ -161,6 +161,8 @@ public void appACLsUpdated(RMApp app, String appViewACLs, public void appAttemptRegistered(RMAppAttempt appAttempt, long registeredTime) { if (publishSystemMetrics) { + ContainerId container = (appAttempt.getMasterContainer() == null) ? null + : appAttempt.getMasterContainer().getId(); dispatcher.getEventHandler().handle( new AppAttemptRegisteredEvent( appAttempt.getAppAttemptId(), @@ -168,7 +170,7 @@ public void appAttemptRegistered(RMAppAttempt appAttempt, appAttempt.getRpcPort(), appAttempt.getTrackingUrl(), appAttempt.getOriginalTrackingUrl(), - appAttempt.getMasterContainer().getId(), + container, registeredTime)); } } @@ -177,6 +179,8 @@ public void appAttemptRegistered(RMAppAttempt appAttempt, public void appAttemptFinished(RMAppAttempt appAttempt, RMAppAttemptState appAttemtpState, RMApp app, long finishedTime) { if (publishSystemMetrics) { + ContainerId container = (appAttempt.getMasterContainer() == null) ? null + : appAttempt.getMasterContainer().getId(); dispatcher.getEventHandler().handle( new AppAttemptFinishedEvent( appAttempt.getAppAttemptId(), @@ -188,7 +192,7 @@ public void appAttemptFinished(RMAppAttempt appAttempt, app.getFinalApplicationStatus(), RMServerUtils.createApplicationAttemptState(appAttemtpState), finishedTime, - appAttempt.getMasterContainer().getId())); + container)); } } @@ -391,9 +395,10 @@ private static TimelineEntity createApplicationEntity( event.getHost()); eventInfo.put(AppAttemptMetricsConstants.RPC_PORT_EVENT_INFO, event.getRpcPort()); - eventInfo.put( - AppAttemptMetricsConstants.MASTER_CONTAINER_EVENT_INFO, - event.getMasterContainerId().toString()); + if (event.getMasterContainerId() != null) { + eventInfo.put(AppAttemptMetricsConstants.MASTER_CONTAINER_EVENT_INFO, + event.getMasterContainerId().toString()); + } tEvent.setEventInfo(eventInfo); entity.addEvent(tEvent); putEntity(entity); @@ -418,9 +423,10 @@ private void publishAppAttemptFinishedEvent(AppAttemptFinishedEvent event) { event.getFinalApplicationStatus().toString()); eventInfo.put(AppAttemptMetricsConstants.STATE_EVENT_INFO, event.getYarnApplicationAttemptState().toString()); - eventInfo.put( - AppAttemptMetricsConstants.MASTER_CONTAINER_EVENT_INFO, - event.getMasterContainerId().toString()); + if (event.getMasterContainerId() != null) { + eventInfo.put(AppAttemptMetricsConstants.MASTER_CONTAINER_EVENT_INFO, + event.getMasterContainerId().toString()); + } tEvent.setEventInfo(eventInfo); entity.addEvent(tEvent); putEntity(entity); diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/TestSystemMetricsPublisher.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/TestSystemMetricsPublisher.java index 4835ade7d12..f2b02bcfad7 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/TestSystemMetricsPublisher.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/metrics/TestSystemMetricsPublisher.java @@ -256,11 +256,31 @@ public void testPublishApplicationMetrics() throws Exception { } } + @Test(timeout = 10000) + public void testPublishAppAttemptMetricsForUnmanagedAM() throws Exception { + ApplicationAttemptId appAttemptId = + ApplicationAttemptId.newInstance(ApplicationId.newInstance(0, 1), 1); + RMAppAttempt appAttempt = createRMAppAttempt(appAttemptId,true); + metricsPublisher.appAttemptRegistered(appAttempt, Integer.MAX_VALUE + 1L); + RMApp app = mock(RMApp.class); + when(app.getFinalApplicationStatus()).thenReturn(FinalApplicationStatus.UNDEFINED); + metricsPublisher.appAttemptFinished(appAttempt, RMAppAttemptState.FINISHED, app, + Integer.MAX_VALUE + 2L); + TimelineEntity entity = null; + do { + entity = + store.getEntity(appAttemptId.toString(), + AppAttemptMetricsConstants.ENTITY_TYPE, + EnumSet.allOf(Field.class)); + // ensure two events are both published before leaving the loop + } while (entity == null || entity.getEvents().size() < 2); + } + @Test(timeout = 10000) public void testPublishAppAttemptMetrics() throws Exception { ApplicationAttemptId appAttemptId = ApplicationAttemptId.newInstance(ApplicationId.newInstance(0, 1), 1); - RMAppAttempt appAttempt = createRMAppAttempt(appAttemptId); + RMAppAttempt appAttempt = createRMAppAttempt(appAttemptId, false); metricsPublisher.appAttemptRegistered(appAttempt, Integer.MAX_VALUE + 1L); RMApp app = mock(RMApp.class); when(app.getFinalApplicationStatus()).thenReturn(FinalApplicationStatus.UNDEFINED); @@ -435,15 +455,17 @@ private static RMApp createRMApp(ApplicationId appId) { } private static RMAppAttempt createRMAppAttempt( - ApplicationAttemptId appAttemptId) { + ApplicationAttemptId appAttemptId, boolean unmanagedAMAttempt) { RMAppAttempt appAttempt = mock(RMAppAttempt.class); when(appAttempt.getAppAttemptId()).thenReturn(appAttemptId); when(appAttempt.getHost()).thenReturn("test host"); when(appAttempt.getRpcPort()).thenReturn(-100); - Container container = mock(Container.class); - when(container.getId()) - .thenReturn(ContainerId.newContainerId(appAttemptId, 1)); - when(appAttempt.getMasterContainer()).thenReturn(container); + if (!unmanagedAMAttempt) { + Container container = mock(Container.class); + when(container.getId()) + .thenReturn(ContainerId.newContainerId(appAttemptId, 1)); + when(appAttempt.getMasterContainer()).thenReturn(container); + } when(appAttempt.getDiagnostics()).thenReturn("test diagnostics info"); when(appAttempt.getTrackingUrl()).thenReturn("test tracking url"); when(appAttempt.getOriginalTrackingUrl()).thenReturn(