From 6772c3f4ddfc14645f26497d346708e910711366 Mon Sep 17 00:00:00 2001 From: Junping Du Date: Wed, 22 Jul 2015 17:52:35 -0700 Subject: [PATCH] YARN-2019. Retrospect on decision of making RM crashed if any exception throw in ZKRMStateStore. Contributed by Jian He. (cherry picked from commit ee98d6354bbbcd0832d3e539ee097f837e5d0e31) --- hadoop-yarn-project/CHANGES.txt | 3 +++ .../hadoop/yarn/conf/YarnConfiguration.java | 11 +++++++++++ .../src/main/resources/yarn-default.xml | 16 ++++++++++++++++ .../resourcemanager/recovery/RMStateStore.java | 9 +++++++-- 4 files changed, 37 insertions(+), 2 deletions(-) diff --git a/hadoop-yarn-project/CHANGES.txt b/hadoop-yarn-project/CHANGES.txt index be74800ae65..7905676ff3c 100644 --- a/hadoop-yarn-project/CHANGES.txt +++ b/hadoop-yarn-project/CHANGES.txt @@ -89,6 +89,9 @@ Release 2.8.0 - UNRELEASED YARN-2003. Support for Application priority : Changes in RM and Capacity Scheduler. (Sunil G via wangda) + YARN-2019. Retrospect on decision of making RM crashed if any exception throw + in ZKRMStateStore. (Jian He via junping_du) + IMPROVEMENTS YARN-644. Basic null check is not performed on passed in arguments before diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java index cdfb393dc1b..7d602a68508 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java @@ -401,6 +401,11 @@ public class YarnConfiguration extends Configuration { public static final String RECOVERY_ENABLED = RM_PREFIX + "recovery.enabled"; public static final boolean DEFAULT_RM_RECOVERY_ENABLED = false; + public static final String YARN_FAIL_FAST = YARN_PREFIX + "fail-fast"; + public static final boolean DEFAULT_YARN_FAIL_FAST = true; + + public static final String RM_FAIL_FAST = RM_PREFIX + "fail-fast"; + @Private public static final String RM_WORK_PRESERVING_RECOVERY_ENABLED = RM_PREFIX + "work-preserving-recovery.enabled"; @@ -2018,6 +2023,12 @@ public class YarnConfiguration extends Configuration { YARN_HTTP_POLICY_DEFAULT)); } + public static boolean shouldRMFailFast(Configuration conf) { + return conf.getBoolean(YarnConfiguration.RM_FAIL_FAST, + conf.getBoolean(YarnConfiguration.YARN_FAIL_FAST, + YarnConfiguration.DEFAULT_YARN_FAIL_FAST)); + } + @Private public static String getClusterId(Configuration conf) { String clusterId = conf.get(YarnConfiguration.RM_CLUSTER_ID); diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml index d586f51ff4a..8b3a3affe9e 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml @@ -323,6 +323,22 @@ false + + Should RM fail fast if it encounters any errors. By defalt, it + points to ${yarn.fail-fast}. Errors include: + 1) exceptions when state-store write/read operations fails. + + yarn.resourcemanager.fail-fast + ${yarn.fail-fast} + + + + Should YARN fail fast if it encounters any errors. + + yarn.fail-fast + true + + Enable RM work preserving recovery. This configuration is private to YARN for experimenting the feature. diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStore.java index 46c2954d301..9b17bf78d57 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStore.java @@ -44,6 +44,7 @@ import org.apache.hadoop.yarn.api.records.ApplicationAttemptId; import org.apache.hadoop.yarn.api.records.ApplicationId; import org.apache.hadoop.yarn.api.records.ApplicationSubmissionContext; import org.apache.hadoop.yarn.api.records.impl.pb.ApplicationSubmissionContextPBImpl; +import org.apache.hadoop.yarn.conf.YarnConfiguration; import org.apache.hadoop.yarn.event.AsyncDispatcher; import org.apache.hadoop.yarn.event.Dispatcher; import org.apache.hadoop.yarn.event.EventHandler; @@ -855,6 +856,7 @@ public abstract class RMStateStore extends AbstractService { * @param failureCause the exception due to which the operation failed */ protected void notifyStoreOperationFailed(Exception failureCause) { + LOG.error("State store operation failed ", failureCause); if (failureCause instanceof StoreFencedException) { updateFencedState(); Thread standByTransitionThread = @@ -862,8 +864,11 @@ public abstract class RMStateStore extends AbstractService { standByTransitionThread.setName("StandByTransitionThread Handler"); standByTransitionThread.start(); } else { - rmDispatcher.getEventHandler().handle( - new RMFatalEvent(RMFatalEventType.STATE_STORE_OP_FAILED, failureCause)); + if (YarnConfiguration.shouldRMFailFast(getConfig())) { + rmDispatcher.getEventHandler().handle( + new RMFatalEvent(RMFatalEventType.STATE_STORE_OP_FAILED, + failureCause)); + } } }