From 9ca2aba9cc65090162b3517b194b5e655ee4a157 Mon Sep 17 00:00:00 2001 From: Jian He Date: Fri, 2 Sep 2016 12:23:57 +0800 Subject: [PATCH] YARN-5601. Make the RM epoch base value configurable. Contributed by Subru Krishnan --- .../hadoop-yarn/dev-support/findbugs-exclude.xml | 5 ++++- .../org/apache/hadoop/yarn/conf/YarnConfiguration.java | 3 +++ .../hadoop/yarn/conf/TestYarnConfigurationFields.java | 2 ++ .../yarn/server/resourcemanager/ResourceManager.java | 7 +++++++ .../resourcemanager/recovery/FileSystemRMStateStore.java | 2 +- .../resourcemanager/recovery/LeveldbRMStateStore.java | 2 +- .../resourcemanager/recovery/MemoryRMStateStore.java | 1 + .../server/resourcemanager/recovery/RMStateStore.java | 4 ++++ .../server/resourcemanager/recovery/ZKRMStateStore.java | 2 +- .../resourcemanager/recovery/RMStateStoreTestBase.java | 8 +++++--- .../resourcemanager/recovery/TestFSRMStateStore.java | 1 + .../resourcemanager/recovery/TestLeveldbRMStateStore.java | 1 + .../resourcemanager/recovery/TestZKRMStateStore.java | 1 + 13 files changed, 32 insertions(+), 7 deletions(-) diff --git a/hadoop-yarn-project/hadoop-yarn/dev-support/findbugs-exclude.xml b/hadoop-yarn-project/hadoop-yarn/dev-support/findbugs-exclude.xml index 2f5451d1c2c..bbd03a9bd4a 100644 --- a/hadoop-yarn-project/hadoop-yarn/dev-support/findbugs-exclude.xml +++ b/hadoop-yarn-project/hadoop-yarn/dev-support/findbugs-exclude.xml @@ -293,7 +293,10 @@ - + + + + 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 46ed221de74..5dcb993dabf 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 @@ -143,6 +143,9 @@ public class YarnConfiguration extends Configuration { public static final String RM_HOSTNAME = RM_PREFIX + "hostname"; + public static final String RM_EPOCH = RM_PREFIX + "epoch"; + public static final long DEFAULT_RM_EPOCH = 0L; + /** The address of the applications manager interface in the RM.*/ public static final String RM_ADDRESS = RM_PREFIX + "address"; diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/test/java/org/apache/hadoop/yarn/conf/TestYarnConfigurationFields.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/test/java/org/apache/hadoop/yarn/conf/TestYarnConfigurationFields.java index 5e0876f6a3e..3f3a06c5fba 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/test/java/org/apache/hadoop/yarn/conf/TestYarnConfigurationFields.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/test/java/org/apache/hadoop/yarn/conf/TestYarnConfigurationFields.java @@ -75,6 +75,8 @@ public class TestYarnConfigurationFields extends TestConfigurationFieldsBase { .add(YarnConfiguration.FEDERATION_FAILOVER_ENABLED); configurationPropsToSkipCompare .add(YarnConfiguration.FEDERATION_STATESTORE_HEARTBEAT_INTERVAL_SECS); + configurationPropsToSkipCompare + .add(YarnConfiguration.RM_EPOCH); // Ignore blacklisting nodes for AM failures feature since it is still a // "work in progress" diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceManager.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceManager.java index 9e78cd7cf15..969188529e8 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceManager.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/ResourceManager.java @@ -772,6 +772,13 @@ public class ResourceManager extends CompositeService implements Recoverable { LOG.error("Failed to load/recover state", e); throw e; } + } else { + if (HAUtil.isFederationEnabled(conf)) { + long epoch = conf.getLong(YarnConfiguration.RM_EPOCH, + YarnConfiguration.DEFAULT_RM_EPOCH); + rmContext.setEpoch(epoch); + LOG.info("Epoch set for Federation: " + epoch); + } } super.serviceStart(); 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/FileSystemRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/FileSystemRMStateStore.java index 95919451a73..7cbeda3a3dd 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/FileSystemRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/FileSystemRMStateStore.java @@ -197,7 +197,7 @@ public class FileSystemRMStateStore extends RMStateStore { @Override public synchronized long getAndIncrementEpoch() throws Exception { Path epochNodePath = getNodePath(rootDirPath, EPOCH_NODE); - long currentEpoch = 0; + long currentEpoch = baseEpoch; FileStatus status = getFileStatusWithRetries(epochNodePath); if (status != null) { // load current epoch 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/LeveldbRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/LeveldbRMStateStore.java index 2ca53db2b34..16ae1d3c19a 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/LeveldbRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/LeveldbRMStateStore.java @@ -255,7 +255,7 @@ public class LeveldbRMStateStore extends RMStateStore { @Override public synchronized long getAndIncrementEpoch() throws Exception { - long currentEpoch = 0; + long currentEpoch = baseEpoch; byte[] dbKeyBytes = bytes(EPOCH_NODE); try { byte[] data = db.get(dbKeyBytes); 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/MemoryRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/MemoryRMStateStore.java index 5f3328b279b..5041000f0ba 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/MemoryRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/MemoryRMStateStore.java @@ -83,6 +83,7 @@ public class MemoryRMStateStore extends RMStateStore { @Override public synchronized void initInternal(Configuration conf) { + epoch = baseEpoch; } @Override 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 d0a8cf5ec4e..465ff64fd52 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 @@ -98,6 +98,7 @@ public abstract class RMStateStore extends AbstractService { "ReservationSystemRoot"; protected static final String VERSION_NODE = "RMVersionNode"; protected static final String EPOCH_NODE = "EpochNode"; + protected long baseEpoch; protected ResourceManager resourceManager; private final ReadLock readLock; private final WriteLock writeLock; @@ -690,6 +691,9 @@ public abstract class RMStateStore extends AbstractService { dispatcher.register(RMStateStoreEventType.class, rmStateStoreEventHandler); dispatcher.setDrainEventsOnStop(); + // read the base epoch value from conf + baseEpoch = conf.getLong(YarnConfiguration.RM_EPOCH, + YarnConfiguration.DEFAULT_RM_EPOCH); initInternal(conf); } 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/ZKRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/ZKRMStateStore.java index 3b986d14a01..1b3b367bb6c 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/ZKRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/ZKRMStateStore.java @@ -438,7 +438,7 @@ public class ZKRMStateStore extends RMStateStore { @Override public synchronized long getAndIncrementEpoch() throws Exception { String epochNodePath = getNodePath(zkRootNodePath, EPOCH_NODE); - long currentEpoch = 0; + long currentEpoch = baseEpoch; if (exists(epochNodePath)) { // load current epoch diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStoreTestBase.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStoreTestBase.java index ca979145308..06a16ffeffe 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStoreTestBase.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/RMStateStoreTestBase.java @@ -91,6 +91,8 @@ public class RMStateStoreTestBase { public static final Log LOG = LogFactory.getLog(RMStateStoreTestBase.class); + protected final long epoch = 10L; + static class TestDispatcher implements Dispatcher, EventHandler { ApplicationAttemptId attemptId; @@ -564,13 +566,13 @@ public class RMStateStoreTestBase { store.setRMDispatcher(new TestDispatcher()); long firstTimeEpoch = store.getAndIncrementEpoch(); - Assert.assertEquals(0, firstTimeEpoch); + Assert.assertEquals(epoch, firstTimeEpoch); long secondTimeEpoch = store.getAndIncrementEpoch(); - Assert.assertEquals(1, secondTimeEpoch); + Assert.assertEquals(epoch + 1, secondTimeEpoch); long thirdTimeEpoch = store.getAndIncrementEpoch(); - Assert.assertEquals(2, thirdTimeEpoch); + Assert.assertEquals(epoch + 2, thirdTimeEpoch); } public void testAppDeletion(RMStateStoreHelper stateStoreHelper) diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestFSRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestFSRMStateStore.java index 5eeb528bf44..0738730aabc 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestFSRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestFSRMStateStore.java @@ -117,6 +117,7 @@ public class TestFSRMStateStore extends RMStateStoreTestBase { conf.setInt(YarnConfiguration.FS_RM_STATE_STORE_NUM_RETRIES, 8); conf.setLong(YarnConfiguration.FS_RM_STATE_STORE_RETRY_INTERVAL_MS, 900L); + conf.setLong(YarnConfiguration.RM_EPOCH, epoch); if (adminCheckEnable) { conf.setBoolean( YarnConfiguration.YARN_INTERMEDIATE_DATA_ENCRYPTION, true); diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestLeveldbRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestLeveldbRMStateStore.java index e3d0f9c22ad..afd0c773bd3 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestLeveldbRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestLeveldbRMStateStore.java @@ -82,6 +82,7 @@ public class TestLeveldbRMStateStore extends RMStateStoreTestBase { @Test(timeout = 60000) public void testEpoch() throws Exception { + conf.setLong(YarnConfiguration.RM_EPOCH, epoch); LeveldbStateStoreTester tester = new LeveldbStateStoreTester(); testEpoch(tester); } diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestZKRMStateStore.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestZKRMStateStore.java index 942e9e88bcf..5ae82391bdf 100644 --- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestZKRMStateStore.java +++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/recovery/TestZKRMStateStore.java @@ -189,6 +189,7 @@ public class TestZKRMStateStore extends RMStateStoreTestBase { conf.set(YarnConfiguration.RM_ZK_ADDRESS, curatorTestingServer.getConnectString()); conf.set(YarnConfiguration.ZK_RM_STATE_STORE_PARENT_PATH, workingZnode); + conf.setLong(YarnConfiguration.RM_EPOCH, epoch); this.store = new TestZKRMStateStoreInternal(conf, workingZnode); return this.store; }