TestDataNodeVolumeFailureToleration.java
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.hdfs.server.datanode;
import static org.apache.hadoop.test.PlatformAssumptions.assumeNotWindows;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertTrue;
import java.io.File;
import java.io.IOException;
import java.util.concurrent.TimeUnit;
import org.apache.hadoop.HadoopIllegalArgumentException;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FileUtil;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hdfs.DFSConfigKeys;
import org.apache.hadoop.hdfs.DFSTestUtil;
import org.apache.hadoop.hdfs.HdfsConfiguration;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.hdfs.server.blockmanagement.DatanodeManager;
import org.apache.hadoop.test.GenericTestUtils;
import org.junit.After;
import org.junit.Before;
import org.junit.Rule;
import org.junit.Test;
import org.junit.rules.Timeout;
/**
* Test the ability of a DN to tolerate volume failures.
*/
public class TestDataNodeVolumeFailureToleration {
private FileSystem fs;
private MiniDFSCluster cluster;
private Configuration conf;
// Sleep at least 3 seconds (a 1s heartbeat plus padding) to allow
// for heartbeats to propagate from the datanodes to the namenode.
final int WAIT_FOR_HEARTBEATS = 3000;
// Wait at least (2 * re-check + 10 * heartbeat) seconds for
// a datanode to be considered dead by the namenode.
final int WAIT_FOR_DEATH = 15000;
// specific the timeout for entire test class
@Rule
public Timeout timeout = new Timeout(120 * 1000);
@Before
public void setUp() throws Exception {
conf = new HdfsConfiguration();
conf.setLong(DFSConfigKeys.DFS_BLOCK_SIZE_KEY, 512L);
/*
* Lower the DN heartbeat, DF rate, and recheck interval to one second
* so state about failures and datanode death propagates faster.
*/
conf.setInt(DFSConfigKeys.DFS_HEARTBEAT_INTERVAL_KEY, 1);
conf.setInt(DFSConfigKeys.DFS_DF_INTERVAL_KEY, 1000);
conf.setInt(DFSConfigKeys.DFS_NAMENODE_HEARTBEAT_RECHECK_INTERVAL_KEY, 1000);
// Allow a single volume failure (there are two volumes)
conf.setInt(DFSConfigKeys.DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY, 1);
cluster = new MiniDFSCluster.Builder(conf).numDataNodes(1).build();
cluster.waitActive();
fs = cluster.getFileSystem();
}
@After
public void tearDown() throws Exception {
if (cluster != null) {
cluster.shutdown();
cluster = null;
}
}
/**
* Test the DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY configuration
* option, ie the DN tolerates a failed-to-use scenario during
* its start-up.
*/
@Test
public void testValidVolumesAtStartup() throws Exception {
assumeNotWindows();
// Make sure no DNs are running.
cluster.shutdownDataNodes();
// Bring up a datanode with two default data dirs, but with one bad one.
conf.setInt(DFSConfigKeys.DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY, 1);
// We use subdirectories 0 and 1 in order to have only a single
// data dir's parent inject a failure.
File tld = new File(MiniDFSCluster.getBaseDirectory(), "badData");
File dataDir1 = new File(tld, "data1");
File dataDir1Actual = new File(dataDir1, "1");
dataDir1Actual.mkdirs();
// Force an IOE to occur on one of the dfs.data.dir.
File dataDir2 = new File(tld, "data2");
prepareDirToFail(dataDir2);
File dataDir2Actual = new File(dataDir2, "2");
// Start one DN, with manually managed DN dir
conf.set(DFSConfigKeys.DFS_DATANODE_DATA_DIR_KEY,
dataDir1Actual.getPath() + "," + dataDir2Actual.getPath());
cluster.startDataNodes(conf, 1, false, null, null);
cluster.waitActive();
try {
assertTrue("The DN should have started up fine.",
cluster.isDataNodeUp());
DataNode dn = cluster.getDataNodes().get(0);
String si = DataNodeTestUtils.getFSDataset(dn).getStorageInfo();
assertTrue("The DN should have started with this directory",
si.contains(dataDir1Actual.getPath()));
assertFalse("The DN shouldn't have a bad directory.",
si.contains(dataDir2Actual.getPath()));
} finally {
cluster.shutdownDataNodes();
FileUtil.chmod(dataDir2.toString(), "755");
}
}
/**
* Test the DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY configuration
* option, ie the DN shuts itself down when the number of failures
* experienced drops below the tolerated amount.
*/
@Test
public void testConfigureMinValidVolumes() throws Exception {
assumeNotWindows();
// Bring up two additional datanodes that need both of their volumes
// functioning in order to stay up.
conf.setInt(DFSConfigKeys.DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY, 0);
conf.setTimeDuration(DFSConfigKeys.DFS_DATANODE_DISK_CHECK_MIN_GAP_KEY,
0, TimeUnit.MILLISECONDS);
cluster.startDataNodes(conf, 2, true, null, null);
cluster.waitActive();
final DatanodeManager dm = cluster.getNamesystem().getBlockManager(
).getDatanodeManager();
long origCapacity = DFSTestUtil.getLiveDatanodeCapacity(dm);
long dnCapacity = DFSTestUtil.getDatanodeCapacity(dm, 0);
// Fail a volume on the 2nd DN
File dn2Vol1 = cluster.getInstanceStorageDir(1, 0);
DataNodeTestUtils.injectDataDirFailure(dn2Vol1);
// Should only get two replicas (the first DN and the 3rd)
Path file1 = new Path("/test1");
DFSTestUtil.createFile(fs, file1, 1024, (short)3, 1L);
DFSTestUtil.waitReplication(fs, file1, (short)2);
// Check that this single failure caused a DN to die.
DFSTestUtil.waitForDatanodeStatus(dm, 2, 1, 0,
origCapacity - (1*dnCapacity), WAIT_FOR_HEARTBEATS);
// If we restore the volume we should still only be able to get
// two replicas since the DN is still considered dead.
DataNodeTestUtils.restoreDataDirFromFailure(dn2Vol1);
Path file2 = new Path("/test2");
DFSTestUtil.createFile(fs, file2, 1024, (short)3, 1L);
DFSTestUtil.waitReplication(fs, file2, (short)2);
}
/**
* Restart the datanodes with a new volume tolerated value.
* @param volTolerated number of dfs data dir failures to tolerate
* @param manageDfsDirs whether the mini cluster should manage data dirs
* @throws IOException
*/
private void restartDatanodes(int volTolerated, boolean manageDfsDirs)
throws IOException {
// Make sure no datanode is running
cluster.shutdownDataNodes();
conf.setInt(DFSConfigKeys.DFS_DATANODE_FAILED_VOLUMES_TOLERATED_KEY, volTolerated);
cluster.startDataNodes(conf, 1, manageDfsDirs, null, null);
cluster.waitActive();
}
/**
* Test for different combination of volume configs and volumes tolerated
* values.
*/
@Test
public void testVolumeAndTolerableConfiguration() throws Exception {
// Check if Block Pool Service exit for an invalid conf value.
testVolumeConfig(-2, 0, false, true);
// Test for one good volume at least
testVolumeConfig(-1, 0, true, true);
testVolumeConfig(-1, 1, true, true);
testVolumeConfig(-1, 2, false, true);
// Ditto if the value is too big.
testVolumeConfig(100, 0, false, true);
// Test for one failed volume
testVolumeConfig(0, 1, false, false);
// Test for one failed volume with 1 tolerable volume
testVolumeConfig(1, 1, true, false);
// Test all good volumes
testVolumeConfig(0, 0, true, false);
// Test all failed volumes
testVolumeConfig(0, 2, false, false);
}
/**
* Tests for a given volumes to be tolerated and volumes failed.
*/
private void testVolumeConfig(int volumesTolerated, int volumesFailed,
boolean expectedBPServiceState, boolean manageDfsDirs)
throws IOException, InterruptedException {
assumeNotWindows();
final int dnIndex = 0;
// Fail the current directory since invalid storage directory perms
// get fixed up automatically on datanode startup.
File[] dirs = {
new File(cluster.getInstanceStorageDir(dnIndex, 0), "current"),
new File(cluster.getInstanceStorageDir(dnIndex, 1), "current") };
try {
for (int i = 0; i < volumesFailed; i++) {
prepareDirToFail(dirs[i]);
}
restartDatanodes(volumesTolerated, manageDfsDirs);
} catch (HadoopIllegalArgumentException e) {
GenericTestUtils.assertExceptionContains("Invalid value configured for "
+ "dfs.datanode.failed.volumes.tolerated", e);
} finally {
for (File dir : dirs) {
FileUtil.chmod(dir.toString(), "755");
}
boolean bpServiceState;
// If the datanode not registered successfully,
// because the invalid value configured for tolerated volumes
if (cluster.getDataNodes().size() == 0) {
bpServiceState = false;
} else {
bpServiceState =
cluster.getDataNodes().get(0)
.isBPServiceAlive(cluster.getNamesystem().getBlockPoolId());
}
assertEquals(expectedBPServiceState, bpServiceState);
}
}
/**
* Prepare directories for a failure, set dir permission to 000
* @param dir
* @throws IOException
* @throws InterruptedException
*/
private void prepareDirToFail(File dir) throws IOException,
InterruptedException {
dir.mkdirs();
assertEquals("Couldn't chmod local vol", 0,
FileUtil.chmod(dir.toString(), "000"));
}
/**
* Test that a volume that is considered failed on startup is seen as
* a failed volume by the NN.
*/
@Test
public void testFailedVolumeOnStartupIsCounted() throws Exception {
assumeNotWindows();
final DatanodeManager dm = cluster.getNamesystem().getBlockManager(
).getDatanodeManager();
long origCapacity = DFSTestUtil.getLiveDatanodeCapacity(dm);
File dir = new File(cluster.getInstanceStorageDir(0, 0), "current");
try {
prepareDirToFail(dir);
restartDatanodes(1, false);
// The cluster is up..
assertEquals(true, cluster.getDataNodes().get(0)
.isBPServiceAlive(cluster.getNamesystem().getBlockPoolId()));
// but there has been a single volume failure
DFSTestUtil.waitForDatanodeStatus(dm, 1, 0, 1,
origCapacity / 2, WAIT_FOR_HEARTBEATS);
} finally {
FileUtil.chmod(dir.toString(), "755");
}
}
}