HDFS-3597. SNN fails to start after DFS upgrade. Contributed by Andy Isaacson.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1363900 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Todd Lipcon 2012-07-20 18:59:17 +00:00
parent 86273d0cf4
commit 75a419f15b
4 changed files with 137 additions and 14 deletions

View File

@ -362,6 +362,8 @@ Release 2.0.1-alpha - UNRELEASED
HDFS-3690. BlockPlacementPolicyDefault incorrectly casts LOG. (eli) HDFS-3690. BlockPlacementPolicyDefault incorrectly casts LOG. (eli)
HDFS-3597. SNN fails to start after DFS upgrade. (Andy Isaacson via todd)
BREAKDOWN OF HDFS-3042 SUBTASKS BREAKDOWN OF HDFS-3042 SUBTASKS
HDFS-2185. HDFS portion of ZK-based FailoverController (todd) HDFS-2185. HDFS portion of ZK-based FailoverController (todd)

View File

@ -113,12 +113,19 @@ public class CheckpointSignature extends StorageInfo
+ blockpoolID ; + blockpoolID ;
} }
boolean storageVersionMatches(StorageInfo si) throws IOException {
return (layoutVersion == si.layoutVersion) && (cTime == si.cTime);
}
boolean isSameCluster(FSImage si) {
return namespaceID == si.getStorage().namespaceID &&
clusterID.equals(si.getClusterID()) &&
blockpoolID.equals(si.getBlockPoolID());
}
void validateStorageInfo(FSImage si) throws IOException { void validateStorageInfo(FSImage si) throws IOException {
if(layoutVersion != si.getStorage().layoutVersion if (!isSameCluster(si)
|| namespaceID != si.getStorage().namespaceID || !storageVersionMatches(si.getStorage())) {
|| cTime != si.getStorage().cTime
|| !clusterID.equals(si.getClusterID())
|| !blockpoolID.equals(si.getBlockPoolID())) {
throw new IOException("Inconsistent checkpoint fields.\n" throw new IOException("Inconsistent checkpoint fields.\n"
+ "LV = " + layoutVersion + " namespaceID = " + namespaceID + "LV = " + layoutVersion + " namespaceID = " + namespaceID
+ " cTime = " + cTime + " cTime = " + cTime

View File

@ -437,18 +437,16 @@ public class SecondaryNameNode implements Runnable {
// Returns a token that would be used to upload the merged image. // Returns a token that would be used to upload the merged image.
CheckpointSignature sig = namenode.rollEditLog(); CheckpointSignature sig = namenode.rollEditLog();
// Make sure we're talking to the same NN! if ((checkpointImage.getNamespaceID() == 0) ||
if (checkpointImage.getNamespaceID() != 0) { (sig.isSameCluster(checkpointImage) &&
// If the image actually has some data, make sure we're talking !sig.storageVersionMatches(checkpointImage.getStorage()))) {
// to the same NN as we did before. // if we're a fresh 2NN, or if we're on the same cluster and our storage
sig.validateStorageInfo(checkpointImage); // needs an upgrade, just take the storage info from the server.
} else {
// if we're a fresh 2NN, just take the storage info from the server
// we first talk to.
dstStorage.setStorageInfo(sig); dstStorage.setStorageInfo(sig);
dstStorage.setClusterID(sig.getClusterID()); dstStorage.setClusterID(sig.getClusterID());
dstStorage.setBlockPoolID(sig.getBlockpoolID()); dstStorage.setBlockPoolID(sig.getBlockpoolID());
} }
sig.validateStorageInfo(checkpointImage);
// error simulation code for junit test // error simulation code for junit test
CheckpointFaultInjector.getInstance().afterSecondaryCallsRollEditLog(); CheckpointFaultInjector.getInstance().afterSecondaryCallsRollEditLog();
@ -703,7 +701,7 @@ public class SecondaryNameNode implements Runnable {
/** /**
* Analyze checkpoint directories. * Analyze checkpoint directories.
* Create directories if they do not exist. * Create directories if they do not exist.
* Recover from an unsuccessful checkpoint is necessary. * Recover from an unsuccessful checkpoint if necessary.
* *
* @throws IOException * @throws IOException
*/ */

View File

@ -0,0 +1,116 @@
/**
* 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.namenode;
import java.io.File;
import java.io.IOException;
import java.util.List;
import org.junit.Test;
import org.junit.Before;
import org.junit.After;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.FileUtil;
import org.apache.hadoop.hdfs.DFSConfigKeys;
import org.apache.hadoop.hdfs.HdfsConfiguration;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import java.util.Properties;
import java.io.FileReader;
import java.io.FileWriter;
import org.junit.Assert;
import org.apache.hadoop.test.GenericTestUtils;
/**
* Regression test for HDFS-3597, SecondaryNameNode upgrade -- when a 2NN
* starts up with an existing directory structure with an old VERSION file, it
* should delete the snapshot and download a new one from the NN.
*/
public class TestSecondaryNameNodeUpgrade {
@Before
public void cleanupCluster() throws IOException {
File hdfsDir = new File(MiniDFSCluster.getBaseDirectory()).getCanonicalFile();
System.out.println("cleanupCluster deleting " + hdfsDir);
if (hdfsDir.exists() && !FileUtil.fullyDelete(hdfsDir)) {
throw new IOException("Could not delete hdfs directory '" + hdfsDir + "'");
}
}
private void doIt(String param, String val) throws IOException {
MiniDFSCluster cluster = null;
FileSystem fs = null;
SecondaryNameNode snn = null;
try {
Configuration conf = new HdfsConfiguration();
cluster = new MiniDFSCluster.Builder(conf).build();
cluster.waitActive();
conf.set(DFSConfigKeys.DFS_NAMENODE_SECONDARY_HTTP_ADDRESS_KEY, "0.0.0.0:0");
snn = new SecondaryNameNode(conf);
fs = cluster.getFileSystem();
fs.mkdirs(new Path("/test/foo"));
snn.doCheckpoint();
List<File> versionFiles = snn.getFSImage().getStorage().getFiles(null, "VERSION");
snn.shutdown();
for (File versionFile : versionFiles) {
System.out.println("Changing '" + param + "' to '" + val + "' in " + versionFile);
FSImageTestUtil.corruptVersionFile(versionFile, param, val);
}
snn = new SecondaryNameNode(conf);
fs.mkdirs(new Path("/test/bar"));
snn.doCheckpoint();
} finally {
if (fs != null) fs.close();
if (cluster != null) cluster.shutdown();
if (snn != null) snn.shutdown();
}
}
@Test
public void testUpgradeLayoutVersionSucceeds() throws IOException {
doIt("layoutVersion", "-39");
}
@Test
public void testChangeNsIDFails() throws IOException {
try {
doIt("namespaceID", "2");
Assert.fail("Should throw InconsistentFSStateException");
} catch(IOException e) {
GenericTestUtils.assertExceptionContains("Inconsistent checkpoint fields", e);
System.out.println("Correctly failed with inconsistent namespaceID: " + e);
}
}
}