big-data/hadoop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

HDFS-11239: [SPS]: Check Mover file ID lease also to determine whether Mover is running. Contributed by Wei Zhou

Browse Source
This commit is contained in:
Rakesh Radhakrishnan 2017-02-17 20:49:38 +05:30 committed by Uma Maheswara Rao Gangumalla
parent e34331c31d
commit 9b15f5418d
4 changed files with 113 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSNamesystem.java
View File

@ -3564,7 +3564,22 @@ void finalizeINodeFileUnderConstruction(String src, INodeFile pendingFile,
BlockInfo getStoredBlock(Block block) {
return blockManager.getStoredBlock(block);
}
@Override
public boolean isFileOpenedForWrite(String path) {
readLock();
try {
INode inode = dir.getINode(path, FSDirectory.DirOp.READ);
INodeFile iNodeFile = INodeFile.valueOf(inode, path);
LeaseManager.Lease lease = leaseManager.getLease(iNodeFile);
return lease != null;
} catch (IOException e) {
return false;
} finally {
readUnlock();
}
}
@Override
public boolean isInSnapshot(long blockCollectionID) {
assert hasReadLock();

7
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/Namesystem.java
View File

@ -45,4 +45,11 @@ public interface Namesystem extends RwLock, SafeMode {
* middle of the starting active services.
*/
boolean inTransitionToActive();
/**
* Check if file is been opened for write purpose.
* @param filePath
* @return true if valid write lease exists, otherwise return false.
*/
boolean isFileOpenedForWrite(String filePath);
}

19
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/StoragePolicySatisfier.java
View File

@ -17,7 +17,6 @@
*/
package org.apache.hadoop.hdfs.server.namenode;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
@ -151,19 +150,8 @@ public boolean isRunning() {
// Return true if a Mover instance is running
private boolean checkIfMoverRunning() {
boolean ret = false;
try {
String moverId = HdfsServerConstants.MOVER_ID_PATH.toString();
INode inode = namesystem.getFSDirectory().getINode(
moverId, FSDirectory.DirOp.READ);
if (inode != null) {
ret = true;
}
} catch (IOException e) {
LOG.info("StoragePolicySatisfier is enabled as no Mover ID file found.");
ret = false;
}
return ret;
String moverId = HdfsServerConstants.MOVER_ID_PATH.toString();
return namesystem.isFileOpenedForWrite(moverId);
}
@Override
@ -177,7 +165,8 @@ public void run() {
this.storageMovementsMonitor.stop();
LOG.error(
"Stopping StoragePolicySatisfier thread " + "as Mover ID file "
+ HdfsServerConstants.MOVER_ID_PATH.toString() + " exists");
+ HdfsServerConstants.MOVER_ID_PATH.toString()
+ " been opened. Maybe a Mover instance is running!");
return;
}
}

108
hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/TestStoragePolicySatisfier.java
View File

@ -29,6 +29,7 @@
import java.util.concurrent.TimeoutException;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.ReconfigurationException;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
@ -97,29 +98,33 @@ public void testWhenStoragePolicySetToCOLD()
try {
createCluster();
// Change policy to COLD
dfs.setStoragePolicy(new Path(file), COLD);
FSNamesystem namesystem = hdfsCluster.getNamesystem();
INode inode = namesystem.getFSDirectory().getINode(file);
StorageType[][] newtypes =
new StorageType[][]{{StorageType.ARCHIVE, StorageType.ARCHIVE},
{StorageType.ARCHIVE, StorageType.ARCHIVE},
{StorageType.ARCHIVE, StorageType.ARCHIVE}};
startAdditionalDNs(config, 3, numOfDatanodes, newtypes,
storagesPerDatanode, capacity, hdfsCluster);
namesystem.getBlockManager().satisfyStoragePolicy(inode.getId());
hdfsCluster.triggerHeartbeats();
// Wait till namenode notified about the block location details
DFSTestUtil.waitExpectedStorageType(
file, StorageType.ARCHIVE, 3, 30000, dfs);
doTestWhenStoragePolicySetToCOLD();
} finally {
shutdownCluster();
}
}
private void doTestWhenStoragePolicySetToCOLD() throws Exception {
// Change policy to COLD
dfs.setStoragePolicy(new Path(file), COLD);
FSNamesystem namesystem = hdfsCluster.getNamesystem();
INode inode = namesystem.getFSDirectory().getINode(file);
StorageType[][] newtypes =
new StorageType[][]{{StorageType.ARCHIVE, StorageType.ARCHIVE},
{StorageType.ARCHIVE, StorageType.ARCHIVE},
{StorageType.ARCHIVE, StorageType.ARCHIVE}};
startAdditionalDNs(config, 3, numOfDatanodes, newtypes,
storagesPerDatanode, capacity, hdfsCluster);
namesystem.getBlockManager().satisfyStoragePolicy(inode.getId());
hdfsCluster.triggerHeartbeats();
// Wait till namenode notified about the block location details
DFSTestUtil.waitExpectedStorageType(
file, StorageType.ARCHIVE, 3, 30000, dfs);
}
@Test(timeout = 300000)
public void testWhenStoragePolicySetToALLSSD()
throws Exception {
@ -500,19 +505,78 @@ public void testWhenNoTargetDatanodeToSatisfyStoragePolicy()
*/
@Test(timeout = 300000)
public void testWhenMoverIsAlreadyRunningBeforeStoragePolicySatisfier()
throws Exception {
boolean running;
FSDataOutputStream out = null;
try {
createCluster();
// Stop SPS
hdfsCluster.getNameNode().reconfigurePropertyImpl(
DFSConfigKeys.DFS_STORAGE_POLICY_SATISFIER_ACTIVATE_KEY, "false");
running = hdfsCluster.getFileSystem()
.getClient().isStoragePolicySatisfierRunning();
Assert.assertFalse("SPS should stopped as configured.", running);
// Simulate the case by creating MOVER_ID file
out = hdfsCluster.getFileSystem().create(
HdfsServerConstants.MOVER_ID_PATH);
// Restart SPS
hdfsCluster.getNameNode().reconfigurePropertyImpl(
DFSConfigKeys.DFS_STORAGE_POLICY_SATISFIER_ACTIVATE_KEY, "true");
running = hdfsCluster.getFileSystem()
.getClient().isStoragePolicySatisfierRunning();
Assert.assertFalse("SPS should not be able to run as file "
+ HdfsServerConstants.MOVER_ID_PATH + " is being hold.", running);
// Simulate Mover exists
out.close();
out = null;
hdfsCluster.getFileSystem().delete(
HdfsServerConstants.MOVER_ID_PATH, true);
// Restart SPS again
hdfsCluster.getNameNode().reconfigurePropertyImpl(
DFSConfigKeys.DFS_STORAGE_POLICY_SATISFIER_ACTIVATE_KEY, "true");
running = hdfsCluster.getFileSystem()
.getClient().isStoragePolicySatisfierRunning();
Assert.assertTrue("SPS should be running as "
+ "Mover already exited", running);
// Check functionality after SPS restart
doTestWhenStoragePolicySetToCOLD();
} catch (ReconfigurationException e) {
throw new IOException("Exception when reconfigure "
+ DFSConfigKeys.DFS_STORAGE_POLICY_SATISFIER_ACTIVATE_KEY, e);
} finally {
if (out != null) {
out.close();
}
hdfsCluster.shutdown();
}
}
/**
* Tests to verify that SPS should be able to start when the Mover ID file
* is not being hold by a Mover. This can be the case when Mover exits
* ungracefully without deleting the ID file from HDFS.
*/
@Test(timeout = 300000)
public void testWhenMoverExitsWithoutDeleteMoverIDFile()
throws IOException {
try {
createCluster();
// Simulate Mover by creating MOVER_ID file
// Simulate the case by creating MOVER_ID file
DFSTestUtil.createFile(hdfsCluster.getFileSystem(),
HdfsServerConstants.MOVER_ID_PATH, 0, (short) 1, 0);
hdfsCluster.restartNameNode(true);
boolean running = hdfsCluster.getFileSystem()
.getClient().isStoragePolicySatisfierRunning();
Assert.assertFalse("SPS should not start "
+ "when a Mover instance is running", running);
Assert.assertTrue("SPS should be running as "
+ "no Mover really running", running);
} finally {
shutdownCluster();
hdfsCluster.shutdown();
}
}