Apache
/
hbase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HBASE-2707 Can't recover from a dead ROOT server if any exceptions happens during log splitting 

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@958153 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Michael Stack 2010-06-26 00:05:17 +00:00
parent ab9f29eafd
commit 7353c3dc0a
8 changed files with 232 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CHANGES.txt
View File

@ -423,6 +423,8 @@ Release 0.21.0 - Unreleased
no progress running YCSB on clean cluster startup
HBASE-2785 TestScannerTimeout.test2772 is flaky
HBASE-2787 PE is confused about flushCommits
HBASE-2707 Can't recover from a dead ROOT server if any exceptions happens
during log splitting
IMPROVEMENTS
HBASE-1760 Cleanup TODOs in HTable

1
src/assembly/bin.xml
View File

@ -28,6 +28,7 @@
<outputDirectory>/</outputDirectory>
<includes>
<include>hbase-${project.version}.jar</include>
<include>hbase-${project.version}-tests.jar</include>
</includes>
</fileSet>
<fileSet>

3
src/main/java/org/apache/hadoop/hbase/master/HMaster.java
View File

@ -504,8 +504,7 @@ public class HMaster extends Thread implements HMasterInterface,
this.serverManager.getServersToServerInfo().keySet().toString());
}
}
final HServerAddress root = this.regionManager.getRootRegionLocation();
switch (this.regionServerOperationQueue.process(root)) {
switch (this.regionServerOperationQueue.process()) {
case FAILED:
// If FAILED op processing, bad. Exit.
break FINISHED;

49
src/main/java/org/apache/hadoop/hbase/master/RegionManager.java
View File

@ -19,32 +19,6 @@
*/
package org.apache.hadoop.hbase.master;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.PathFilter;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.HMsg;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.HServerAddress;
import org.apache.hadoop.hbase.HServerInfo;
import org.apache.hadoop.hbase.HServerLoad;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.executor.RegionTransitionEventData;
import org.apache.hadoop.hbase.executor.HBaseEventHandler;
import org.apache.hadoop.hbase.executor.HBaseEventHandler.HBaseEventType;
import org.apache.hadoop.hbase.ipc.HRegionInterface;
import org.apache.hadoop.hbase.regionserver.HRegion;
import org.apache.hadoop.hbase.regionserver.wal.HLog;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.Pair;
import org.apache.hadoop.hbase.util.Threads;
import org.apache.hadoop.hbase.util.Writables;
import org.apache.hadoop.hbase.zookeeper.ZooKeeperWrapper;
import org.apache.hadoop.io.WritableUtils;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collections;
@ -61,6 +35,29 @@ import java.util.concurrent.ConcurrentSkipListMap;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicReference;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.PathFilter;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.HMsg;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.HServerAddress;
import org.apache.hadoop.hbase.HServerInfo;
import org.apache.hadoop.hbase.HServerLoad;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.executor.RegionTransitionEventData;
import org.apache.hadoop.hbase.executor.HBaseEventHandler.HBaseEventType;
import org.apache.hadoop.hbase.ipc.HRegionInterface;
import org.apache.hadoop.hbase.regionserver.HRegion;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.Pair;
import org.apache.hadoop.hbase.util.Threads;
import org.apache.hadoop.hbase.util.Writables;
import org.apache.hadoop.hbase.zookeeper.ZooKeeperWrapper;
/**
* Class to manage assigning regions to servers, state of root and meta, etc.
*/

5
src/main/java/org/apache/hadoop/hbase/master/RegionServerOperationQueue.java
View File

@ -111,17 +111,16 @@ public class RegionServerOperationQueue {
/**
* Try to get an operation off of the queue and process it.
* @param rootRegionLocation Location of the root region.
* @return {@link ProcessingResultCode#PROCESSED},
* {@link ProcessingResultCode#REQUEUED},
* {@link ProcessingResultCode#REQUEUED_BUT_PROBLEM}
*/
public synchronized ProcessingResultCode process(final HServerAddress rootRegionLocation) {
public synchronized ProcessingResultCode process() {
RegionServerOperation op = null;
// Only process the delayed queue if root region is online. If offline,
// the operation to put it online is probably in the toDoQueue. Process
// it first.
if (rootRegionLocation != null) {
if (toDoQueue.isEmpty()) {
op = delayedToDoQueue.poll();
}
if (op == null) {

34
src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java
View File

@ -50,6 +50,7 @@ import org.apache.hadoop.hbase.master.HMaster;
import org.apache.hadoop.hbase.regionserver.HRegionServer;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.FSUtils;
import org.apache.hadoop.hbase.util.Threads;
import org.apache.hadoop.hbase.util.Writables;
import org.apache.hadoop.hbase.zookeeper.ZooKeeperWrapper;
import org.apache.hadoop.hdfs.MiniDFSCluster;
@ -920,4 +921,35 @@ public class HBaseTestingUtility {
LOG.info("Could not set max recovery field", e);
}
}
}
/**
* Wait until <code>countOfRegion</code> in .META. have a non-empty
* info:server. This means all regions have been deployed, master has been
* informed and updated .META. with the regions deployed server.
* @param conf Configuration
* @param countOfRegions How many regions in .META.
* @throws IOException
*/
public void waitUntilAllRegionsAssigned(final int countOfRegions)
throws IOException {
HTable meta = new HTable(getConfiguration(), HConstants.META_TABLE_NAME);
while (true) {
int rows = 0;
Scan scan = new Scan();
scan.addColumn(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
ResultScanner s = meta.getScanner(scan);
for (Result r = null; (r = s.next()) != null;) {
byte [] b =
r.getValue(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
if (b == null || b.length <= 0) break;
rows++;
}
s.close();
// If I get to here and all rows have a Server, then all have been assigned.
if (rows == countOfRegions) break;
LOG.info("Found=" + rows);
Threads.sleep(1000);
}
}
}

32
src/test/java/org/apache/hadoop/hbase/master/TestMasterTransitions.java
View File

@ -84,7 +84,7 @@ public class TestMasterTransitions {
TEST_UTIL.createTable(Bytes.toBytes(TABLENAME), FAMILIES);
HTable t = new HTable(TEST_UTIL.getConfiguration(), TABLENAME);
int countOfRegions = TEST_UTIL.createMultiRegions(t, getTestFamily());
waitUntilAllRegionsAssigned(countOfRegions);
TEST_UTIL.waitUntilAllRegionsAssigned(countOfRegions);
addToEachStartKey(countOfRegions);
}
@ -456,36 +456,6 @@ public class TestMasterTransitions {
assertTrue((t.get(g)).size() > 0);
}
/*
* Wait until all rows in .META. have a non-empty info:server. This means
* all regions have been deployed, master has been informed and updated
* .META. with the regions deployed server.
* @param countOfRegions How many regions in .META.
* @throws IOException
*/
private static void waitUntilAllRegionsAssigned(final int countOfRegions)
throws IOException {
HTable meta = new HTable(TEST_UTIL.getConfiguration(),
HConstants.META_TABLE_NAME);
while (true) {
int rows = 0;
Scan scan = new Scan();
scan.addColumn(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
ResultScanner s = meta.getScanner(scan);
for (Result r = null; (r = s.next()) != null;) {
byte [] b =
r.getValue(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
if (b == null || b.length <= 0) break;
rows++;
}
s.close();
// If I get to here and all rows have a Server, then all have been assigned.
if (rows == countOfRegions) break;
LOG.info("Found=" + rows);
Threads.sleep(1000);
}
}
/*
* @return Count of regions in meta table.
* @throws IOException

169
src/test/java/org/apache/hadoop/hbase/master/TestROOTAssignment.java Normal file
View File

@ -0,0 +1,169 @@
/**
* Copyright 2010 The Apache Software Foundation
*
* 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.hbase.master;
import java.io.IOException;
import java.util.Set;
import java.util.concurrent.CopyOnWriteArraySet;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.hbase.HBaseTestingUtility;
import org.apache.hadoop.hbase.HMsg;
import org.apache.hadoop.hbase.HServerAddress;
import org.apache.hadoop.hbase.HServerInfo;
import org.apache.hadoop.hbase.MiniHBaseCluster;
import org.apache.hadoop.hbase.client.HTable;
import org.apache.hadoop.hbase.regionserver.HRegionServer;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.Threads;
import org.junit.AfterClass;
import org.junit.Assert;
import org.junit.Before;
import org.junit.BeforeClass;
import org.junit.Test;
/**
* Test issues assigning ROOT.
*/
public class TestROOTAssignment {
private static final Log LOG = LogFactory.getLog(TestROOTAssignment.class);
private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
private static final byte [] TABLENAME = Bytes.toBytes("root_assignments");
private static final byte [][] FAMILIES =
new byte [][] {Bytes.toBytes("family")};
/**
* Start up a mini cluster and put a small table of many empty regions into it.
* @throws Exception
*/
@BeforeClass public static void beforeAllTests() throws Exception {
TEST_UTIL.getConfiguration().setInt("hbase.regions.percheckin", 2);
// Start a cluster of two regionservers.
TEST_UTIL.startMiniCluster(2);
// Create a table of three families. This will assign a region.
TEST_UTIL.createTable(TABLENAME, FAMILIES);
HTable t = new HTable(TEST_UTIL.getConfiguration(), TABLENAME);
int countOfRegions = TEST_UTIL.createMultiRegions(t, FAMILIES[0]);
TEST_UTIL.waitUntilAllRegionsAssigned(countOfRegions);
HTable table = new HTable(TEST_UTIL.getConfiguration(), TABLENAME);
TEST_UTIL.loadTable(table, FAMILIES[0]);
table.close();
}
@AfterClass public static void afterAllTests() throws IOException {
TEST_UTIL.shutdownMiniCluster();
}
@Before public void setup() throws IOException {
TEST_UTIL.ensureSomeRegionServersAvailable(2);
}
/**
* Interrupt processing of server shutdown so it gets put on delay queue.
*/
static class PostponeShutdownProcessing implements RegionServerOperationListener {
// Map of what we've delayed so we don't do do repeated delays.
private final Set<RegionServerOperation> postponed =
new CopyOnWriteArraySet<RegionServerOperation>();
private boolean done = false;
private final HServerAddress rootServerAddress;
private final HMaster master;
PostponeShutdownProcessing(final HMaster master,
final HServerAddress rootServerAddress) {
this.master = master;
this.rootServerAddress = rootServerAddress;
}
@Override
public boolean process(final RegionServerOperation op) throws IOException {
// If a regionserver shutdown and its of the root server, then we want to
// delay the processing of the shutdown
boolean result = true;
if (op instanceof ProcessServerShutdown) {
ProcessServerShutdown pss = (ProcessServerShutdown)op;
if (pss.getDeadServerAddress().equals(this.rootServerAddress)) {
// Don't postpone more than once.
if (!this.postponed.contains(pss)) {
this.postponed.add(pss);
Assert.assertNull(this.master.getRegionManager().getRootRegionLocation());
pss.setDelay(1 * 1000);
// Return false. This will add this op to the delayed queue.
result = false;
}
}
}
return result;
}
@Override
public boolean process(HServerInfo serverInfo, HMsg incomingMsg) {
return true;
}
@Override
public void processed(RegionServerOperation op) {
if (op instanceof ProcessServerShutdown) {
ProcessServerShutdown pss = (ProcessServerShutdown)op;
if (pss.getDeadServerAddress().equals(this.rootServerAddress)) {
this.done = true;
}
}
}
public boolean isDone() {
return this.done;
}
}
/**
* If the split of the log for the regionserver hosting ROOT doesn't go off
* smoothly, if the process server shutdown gets added to the delayed queue
* of events to process, then ROOT was not being allocated, ever.
* @see <a href="https://issues.apache.org/jira/browse/HBASE-2707">HBASE-2707</a>
*/
@Test (timeout=300000) public void testROOTDeployedThoughProblemSplittingLog()
throws Exception {
LOG.info("Running testROOTDeployedThoughProblemSplittingLog");
MiniHBaseCluster cluster = TEST_UTIL.getHBaseCluster();
final HMaster master = cluster.getMaster();
byte [] rootRegion = Bytes.toBytes("-ROOT-,,0");
int rootIndex = cluster.getServerWith(rootRegion);
final HRegionServer rootHRS = cluster.getRegionServer(rootIndex);
// Add our RegionServerOperationsListener
PostponeShutdownProcessing listener = new PostponeShutdownProcessing(master,
rootHRS.getHServerInfo().getServerAddress());
master.getRegionServerOperationQueue().
registerRegionServerOperationListener(listener);
try {
// Now close the server carrying meta.
cluster.abortRegionServer(rootIndex);
// Wait for processing of the shutdown server.
while(!listener.isDone()) Threads.sleep(100);
master.getRegionManager().waitForRootRegionLocation();
} finally {
master.getRegionServerOperationQueue().
unregisterRegionServerOperationListener(listener);
}
}
}