- 浏览: 96445 次
- 性别:
- 来自: 北京
最新评论
-
leibnitz:
这么早就研究源码已经是难得了.你说的第二点能这样说吗:a.会依 ...
Hbase的Region Compact算法实现分析 -
uestzengting:
朝阳之辉 写道你的想法是保证rowkey唯一是吧,我的想法是r ...
HBase bulkload的一个bug定位 -
朝阳之辉:
你的想法是保证rowkey唯一是吧,我的想法是rowkey是可 ...
HBase bulkload的一个bug定位 -
uestzengting:
朝阳之辉 写道你的version是用的时间戳吧,在数据量非常大 ...
HBase bulkload的一个bug定位 -
朝阳之辉:
你的version是用的时间戳吧,在数据量非常大的情况下,怎样 ...
HBase bulkload的一个bug定位
随着Hbase里删除的进行,有些Region的数据会越来越少,而HBase不会主动去回收这些Region,因此会造成Region越来越多。HBase里提供了一个工具类HMerge,直接拿过来用却并不能运行,按照自已对HBase的理解对HMerge稍做修改,使其可以运行。运行时不需要Disable表,但需要注意的是在运行时如果同时往该表里put数据,可能会有问题,改后的代码如下:
/**
* Copyright 2009 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 hbase;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Random;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.HTableDescriptor;
import org.apache.hadoop.hbase.KeyValue;
import org.apache.hadoop.hbase.RemoteExceptionHandler;
import org.apache.hadoop.hbase.client.Delete;
import org.apache.hadoop.hbase.client.HBaseAdmin;
import org.apache.hadoop.hbase.client.HConnection;
import org.apache.hadoop.hbase.client.HConnectionManager;
import org.apache.hadoop.hbase.client.HTable;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.client.Result;
import org.apache.hadoop.hbase.client.ResultScanner;
import org.apache.hadoop.hbase.client.Scan;
import org.apache.hadoop.hbase.regionserver.HRegion;
import org.apache.hadoop.hbase.regionserver.InternalScanner;
import org.apache.hadoop.hbase.regionserver.wal.HLog;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.Writables;
/**
* A non-instantiable class that has a static method capable of compacting
* a table by merging adjacent regions.
*/
public class HMerge {
static final Log LOG = LogFactory.getLog(HMerge.class);
static final Random rand = new Random();
/*
* Not instantiable
*/
private HMerge() {
super();
}
/**
* Scans the table and merges two adjacent regions if they are small. This
* only happens when a lot of rows are deleted.
*
* When merging the META region, the HBase instance must be offline.
* When merging a normal table, the HBase instance must be online, but the
* table must be disabled.
*
* @param conf - configuration object for HBase
* @param fs - FileSystem where regions reside
* @param tableName - Table to be compacted
* @throws IOException
*/
public static void merge(Configuration conf, FileSystem fs, final byte[] tableName)
throws IOException {
merge(conf, fs, tableName, true);
}
/**
* Scans the table and merges two adjacent regions if they are small. This
* only happens when a lot of rows are deleted.
*
* When merging the META region, the HBase instance must be offline.
* When merging a normal table, the HBase instance must be online, but the
* table must be disabled.
*
* @param conf - configuration object for HBase
* @param fs - FileSystem where regions reside
* @param tableName - Table to be compacted
* @param testMasterRunning True if we are to verify master is down before
* running merge
* @throws IOException
*/
public static void merge(Configuration conf, FileSystem fs, final byte[] tableName,
final boolean testMasterRunning) throws IOException {
boolean masterIsRunning = false;
if (testMasterRunning) {
HConnection connection = HConnectionManager.getConnection(conf);
masterIsRunning = connection.isMasterRunning();
}
HConnectionManager.deleteConnection(conf, true);
if (Bytes.equals(tableName, HConstants.META_TABLE_NAME)) {
if (masterIsRunning) {
throw new IllegalStateException("Can not compact META table if instance is on-line");
}
new OfflineMerger(conf, fs).process();
} else {
if (!masterIsRunning) {
throw new IllegalStateException(
"HBase instance must be running to merge a normal table");
}
// HBaseAdmin admin = new HBaseAdmin(conf);
// if (!admin.isTableDisabled(tableName)) {
// throw new TableNotDisabledException(tableName);
// }
new OnlineMerger(conf, fs, tableName).process();
}
}
private static abstract class Merger {
protected final Configuration conf;
protected final FileSystem fs;
protected final Path tabledir;
protected final HLog hlog;
private final long maxFilesize;
protected Merger(Configuration conf, FileSystem fs, final byte[] tableName)
throws IOException {
this.conf = conf;
this.fs = fs;
this.maxFilesize = conf.getLong("hbase.hregion.max.filesize",
HConstants.DEFAULT_MAX_FILE_SIZE);
this.tabledir = new Path(fs.makeQualified(new Path(conf.get(HConstants.HBASE_DIR))),
Bytes.toString(tableName));
Path logdir = new Path(tabledir, "merge_" + System.currentTimeMillis()
+ HConstants.HREGION_LOGDIR_NAME);
Path oldLogDir = new Path(tabledir, HConstants.HREGION_OLDLOGDIR_NAME);
this.hlog = new HLog(fs, logdir, oldLogDir, conf);
}
void process() throws IOException {
try {
for (HRegionInfo[] regionsToMerge = next(); regionsToMerge != null; regionsToMerge = next()) {
if (!merge(regionsToMerge)) {
return;
}
}
} finally {
try {
hlog.closeAndDelete();
} catch (IOException e) {
LOG.error(e);
}
}
}
protected boolean merge(final HRegionInfo[] info) throws IOException {
if (info.length < 2) {
LOG.info("only one region - nothing to merge");
return false;
}
HRegion currentRegion = null;
long currentSize = 0;
HRegion nextRegion = null;
long nextSize = 0;
for (int i = 0; i < info.length - 1; i++) {
if (currentRegion == null) {
currentRegion = HRegion.newHRegion(tabledir, hlog, fs, conf, info[i], null);
currentRegion.initialize();
currentSize = currentRegion.getLargestHStoreSize();
}
nextRegion = HRegion.newHRegion(tabledir, hlog, fs, conf, info[i + 1], null);
nextRegion.initialize();
nextSize = nextRegion.getLargestHStoreSize();
if ((currentSize + nextSize) <= (maxFilesize / 2)) {
// We merge two adjacent regions if their total size is less than
// one half of the desired maximum size
LOG.info("Merging regions " + currentRegion.getRegionNameAsString() + " and "
+ nextRegion.getRegionNameAsString());
HRegion mergedRegion = HRegion.mergeAdjacent(currentRegion, nextRegion);
updateMeta(currentRegion.getRegionName(), nextRegion.getRegionName(),
mergedRegion);
break;
}
LOG.info("not merging regions " + Bytes.toString(currentRegion.getRegionName())
+ " and " + Bytes.toString(nextRegion.getRegionName()));
currentRegion.close();
currentRegion = nextRegion;
currentSize = nextSize;
}
if (currentRegion != null) {
currentRegion.close();
}
return true;
}
protected abstract HRegionInfo[] next() throws IOException;
protected abstract void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException;
}
/** Instantiated to compact a normal user table */
private static class OnlineMerger extends Merger {
private final byte[] tableName;
private final HTable table;
private final ResultScanner metaScanner;
private HRegionInfo latestRegion;
OnlineMerger(Configuration conf, FileSystem fs, final byte[] tableName) throws IOException {
super(conf, fs, tableName);
this.tableName = tableName;
this.table = new HTable(conf, HConstants.META_TABLE_NAME);
this.metaScanner = table.getScanner(HConstants.CATALOG_FAMILY,
HConstants.REGIONINFO_QUALIFIER);
this.latestRegion = null;
}
private HRegionInfo nextRegion() throws IOException {
try {
HRegionInfo results = getMetaRow();
return results;
} catch (IOException e) {
e = RemoteExceptionHandler.checkIOException(e);
LOG.error("meta scanner error", e);
metaScanner.close();
throw e;
}
}
/*
* Check current row has a HRegionInfo. Skip to next row if HRI is empty.
* @return A Map of the row content else null if we are off the end.
* @throws IOException
*/
private HRegionInfo getMetaRow() throws IOException {
Result currentRow = metaScanner.next();
boolean found = false;
HRegionInfo region = null;
while (currentRow != null) {
LOG.info("Row: <" + Bytes.toString(currentRow.getRow()) + ">");
byte[] regionInfoValue = currentRow.getValue(HConstants.CATALOG_FAMILY,
HConstants.REGIONINFO_QUALIFIER);
if (regionInfoValue == null || regionInfoValue.length == 0) {
currentRow = metaScanner.next();
continue;
} else {
region = Writables.getHRegionInfo(regionInfoValue);
if (!Bytes.equals(region.getTableDesc().getName(), this.tableName)) {
currentRow = metaScanner.next();
continue;
}
}
found = true;
break;
}
return found ? region : null;
}
@Override
protected HRegionInfo[] next() throws IOException {
List<HRegionInfo> regions = new ArrayList<HRegionInfo>();
if (latestRegion == null) {
latestRegion = nextRegion();
}
if (latestRegion != null) {
regions.add(latestRegion);
}
latestRegion = nextRegion();
if (latestRegion != null) {
regions.add(latestRegion);
}
return regions.toArray(new HRegionInfo[regions.size()]);
}
@Override
protected void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException {
byte[][] regionsToDelete = { oldRegion1, oldRegion2 };
for (int r = 0; r < regionsToDelete.length; r++) {
if (Bytes.equals(regionsToDelete[r], latestRegion.getRegionName())) {
latestRegion = null;
}
Delete delete = new Delete(regionsToDelete[r]);
table.delete(delete);
LOG.info("updated columns in row: " + Bytes.toString(regionsToDelete[r]));
}
Put put = new Put(newRegion.getRegionName());
put.add(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER, Writables
.getBytes(newRegion.getRegionInfo()));
table.put(put);
HBaseAdmin admin = new HBaseAdmin(conf);
admin.getMaster().assign(newRegion.getRegionName(), true);
LOG.info("updated columns in row: " + Bytes.toString(newRegion.getRegionName()));
}
}
/** Instantiated to compact the meta region */
private static class OfflineMerger extends Merger {
private final List<HRegionInfo> metaRegions = new ArrayList<HRegionInfo>();
private final HRegion root;
OfflineMerger(Configuration conf, FileSystem fs) throws IOException {
super(conf, fs, HConstants.META_TABLE_NAME);
Path rootTableDir = HTableDescriptor.getTableDir(fs.makeQualified(new Path(conf
.get(HConstants.HBASE_DIR))), HConstants.ROOT_TABLE_NAME);
// Scan root region to find all the meta regions
root = HRegion.newHRegion(rootTableDir, hlog, fs, conf, HRegionInfo.ROOT_REGIONINFO,
null);
root.initialize();
Scan scan = new Scan();
scan.addColumn(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER);
InternalScanner rootScanner = root.getScanner(scan);
try {
List<KeyValue> results = new ArrayList<KeyValue>();
while (rootScanner.next(results)) {
for (KeyValue kv : results) {
HRegionInfo info = Writables.getHRegionInfoOrNull(kv.getValue());
if (info != null) {
metaRegions.add(info);
}
}
}
} finally {
rootScanner.close();
try {
root.close();
} catch (IOException e) {
LOG.error(e);
}
}
}
@Override
protected HRegionInfo[] next() {
HRegionInfo[] results = null;
if (metaRegions.size() > 0) {
results = metaRegions.toArray(new HRegionInfo[metaRegions.size()]);
metaRegions.clear();
}
return results;
}
@Override
protected void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException {
byte[][] regionsToDelete = { oldRegion1, oldRegion2 };
for (int r = 0; r < regionsToDelete.length; r++) {
Delete delete = new Delete(regionsToDelete[r]);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.STARTCODE_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SPLITA_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SPLITB_QUALIFIER);
root.delete(delete, null, true);
LOG.info("updated columns in row: " + Bytes.toString(regionsToDelete[r]));
}
HRegionInfo newInfo = newRegion.getRegionInfo();
newInfo.setOffline(true);
Put put = new Put(newRegion.getRegionName());
put.add(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER, Writables
.getBytes(newInfo));
root.put(put);
LOG.info("updated columns in row: " + Bytes.toString(newRegion.getRegionName()));
}
}
}
/**
* Copyright 2009 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 hbase;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Random;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.HTableDescriptor;
import org.apache.hadoop.hbase.KeyValue;
import org.apache.hadoop.hbase.RemoteExceptionHandler;
import org.apache.hadoop.hbase.client.Delete;
import org.apache.hadoop.hbase.client.HBaseAdmin;
import org.apache.hadoop.hbase.client.HConnection;
import org.apache.hadoop.hbase.client.HConnectionManager;
import org.apache.hadoop.hbase.client.HTable;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.client.Result;
import org.apache.hadoop.hbase.client.ResultScanner;
import org.apache.hadoop.hbase.client.Scan;
import org.apache.hadoop.hbase.regionserver.HRegion;
import org.apache.hadoop.hbase.regionserver.InternalScanner;
import org.apache.hadoop.hbase.regionserver.wal.HLog;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.Writables;
/**
* A non-instantiable class that has a static method capable of compacting
* a table by merging adjacent regions.
*/
public class HMerge {
static final Log LOG = LogFactory.getLog(HMerge.class);
static final Random rand = new Random();
/*
* Not instantiable
*/
private HMerge() {
super();
}
/**
* Scans the table and merges two adjacent regions if they are small. This
* only happens when a lot of rows are deleted.
*
* When merging the META region, the HBase instance must be offline.
* When merging a normal table, the HBase instance must be online, but the
* table must be disabled.
*
* @param conf - configuration object for HBase
* @param fs - FileSystem where regions reside
* @param tableName - Table to be compacted
* @throws IOException
*/
public static void merge(Configuration conf, FileSystem fs, final byte[] tableName)
throws IOException {
merge(conf, fs, tableName, true);
}
/**
* Scans the table and merges two adjacent regions if they are small. This
* only happens when a lot of rows are deleted.
*
* When merging the META region, the HBase instance must be offline.
* When merging a normal table, the HBase instance must be online, but the
* table must be disabled.
*
* @param conf - configuration object for HBase
* @param fs - FileSystem where regions reside
* @param tableName - Table to be compacted
* @param testMasterRunning True if we are to verify master is down before
* running merge
* @throws IOException
*/
public static void merge(Configuration conf, FileSystem fs, final byte[] tableName,
final boolean testMasterRunning) throws IOException {
boolean masterIsRunning = false;
if (testMasterRunning) {
HConnection connection = HConnectionManager.getConnection(conf);
masterIsRunning = connection.isMasterRunning();
}
HConnectionManager.deleteConnection(conf, true);
if (Bytes.equals(tableName, HConstants.META_TABLE_NAME)) {
if (masterIsRunning) {
throw new IllegalStateException("Can not compact META table if instance is on-line");
}
new OfflineMerger(conf, fs).process();
} else {
if (!masterIsRunning) {
throw new IllegalStateException(
"HBase instance must be running to merge a normal table");
}
// HBaseAdmin admin = new HBaseAdmin(conf);
// if (!admin.isTableDisabled(tableName)) {
// throw new TableNotDisabledException(tableName);
// }
new OnlineMerger(conf, fs, tableName).process();
}
}
private static abstract class Merger {
protected final Configuration conf;
protected final FileSystem fs;
protected final Path tabledir;
protected final HLog hlog;
private final long maxFilesize;
protected Merger(Configuration conf, FileSystem fs, final byte[] tableName)
throws IOException {
this.conf = conf;
this.fs = fs;
this.maxFilesize = conf.getLong("hbase.hregion.max.filesize",
HConstants.DEFAULT_MAX_FILE_SIZE);
this.tabledir = new Path(fs.makeQualified(new Path(conf.get(HConstants.HBASE_DIR))),
Bytes.toString(tableName));
Path logdir = new Path(tabledir, "merge_" + System.currentTimeMillis()
+ HConstants.HREGION_LOGDIR_NAME);
Path oldLogDir = new Path(tabledir, HConstants.HREGION_OLDLOGDIR_NAME);
this.hlog = new HLog(fs, logdir, oldLogDir, conf);
}
void process() throws IOException {
try {
for (HRegionInfo[] regionsToMerge = next(); regionsToMerge != null; regionsToMerge = next()) {
if (!merge(regionsToMerge)) {
return;
}
}
} finally {
try {
hlog.closeAndDelete();
} catch (IOException e) {
LOG.error(e);
}
}
}
protected boolean merge(final HRegionInfo[] info) throws IOException {
if (info.length < 2) {
LOG.info("only one region - nothing to merge");
return false;
}
HRegion currentRegion = null;
long currentSize = 0;
HRegion nextRegion = null;
long nextSize = 0;
for (int i = 0; i < info.length - 1; i++) {
if (currentRegion == null) {
currentRegion = HRegion.newHRegion(tabledir, hlog, fs, conf, info[i], null);
currentRegion.initialize();
currentSize = currentRegion.getLargestHStoreSize();
}
nextRegion = HRegion.newHRegion(tabledir, hlog, fs, conf, info[i + 1], null);
nextRegion.initialize();
nextSize = nextRegion.getLargestHStoreSize();
if ((currentSize + nextSize) <= (maxFilesize / 2)) {
// We merge two adjacent regions if their total size is less than
// one half of the desired maximum size
LOG.info("Merging regions " + currentRegion.getRegionNameAsString() + " and "
+ nextRegion.getRegionNameAsString());
HRegion mergedRegion = HRegion.mergeAdjacent(currentRegion, nextRegion);
updateMeta(currentRegion.getRegionName(), nextRegion.getRegionName(),
mergedRegion);
break;
}
LOG.info("not merging regions " + Bytes.toString(currentRegion.getRegionName())
+ " and " + Bytes.toString(nextRegion.getRegionName()));
currentRegion.close();
currentRegion = nextRegion;
currentSize = nextSize;
}
if (currentRegion != null) {
currentRegion.close();
}
return true;
}
protected abstract HRegionInfo[] next() throws IOException;
protected abstract void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException;
}
/** Instantiated to compact a normal user table */
private static class OnlineMerger extends Merger {
private final byte[] tableName;
private final HTable table;
private final ResultScanner metaScanner;
private HRegionInfo latestRegion;
OnlineMerger(Configuration conf, FileSystem fs, final byte[] tableName) throws IOException {
super(conf, fs, tableName);
this.tableName = tableName;
this.table = new HTable(conf, HConstants.META_TABLE_NAME);
this.metaScanner = table.getScanner(HConstants.CATALOG_FAMILY,
HConstants.REGIONINFO_QUALIFIER);
this.latestRegion = null;
}
private HRegionInfo nextRegion() throws IOException {
try {
HRegionInfo results = getMetaRow();
return results;
} catch (IOException e) {
e = RemoteExceptionHandler.checkIOException(e);
LOG.error("meta scanner error", e);
metaScanner.close();
throw e;
}
}
/*
* Check current row has a HRegionInfo. Skip to next row if HRI is empty.
* @return A Map of the row content else null if we are off the end.
* @throws IOException
*/
private HRegionInfo getMetaRow() throws IOException {
Result currentRow = metaScanner.next();
boolean found = false;
HRegionInfo region = null;
while (currentRow != null) {
LOG.info("Row: <" + Bytes.toString(currentRow.getRow()) + ">");
byte[] regionInfoValue = currentRow.getValue(HConstants.CATALOG_FAMILY,
HConstants.REGIONINFO_QUALIFIER);
if (regionInfoValue == null || regionInfoValue.length == 0) {
currentRow = metaScanner.next();
continue;
} else {
region = Writables.getHRegionInfo(regionInfoValue);
if (!Bytes.equals(region.getTableDesc().getName(), this.tableName)) {
currentRow = metaScanner.next();
continue;
}
}
found = true;
break;
}
return found ? region : null;
}
@Override
protected HRegionInfo[] next() throws IOException {
List<HRegionInfo> regions = new ArrayList<HRegionInfo>();
if (latestRegion == null) {
latestRegion = nextRegion();
}
if (latestRegion != null) {
regions.add(latestRegion);
}
latestRegion = nextRegion();
if (latestRegion != null) {
regions.add(latestRegion);
}
return regions.toArray(new HRegionInfo[regions.size()]);
}
@Override
protected void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException {
byte[][] regionsToDelete = { oldRegion1, oldRegion2 };
for (int r = 0; r < regionsToDelete.length; r++) {
if (Bytes.equals(regionsToDelete[r], latestRegion.getRegionName())) {
latestRegion = null;
}
Delete delete = new Delete(regionsToDelete[r]);
table.delete(delete);
LOG.info("updated columns in row: " + Bytes.toString(regionsToDelete[r]));
}
Put put = new Put(newRegion.getRegionName());
put.add(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER, Writables
.getBytes(newRegion.getRegionInfo()));
table.put(put);
HBaseAdmin admin = new HBaseAdmin(conf);
admin.getMaster().assign(newRegion.getRegionName(), true);
LOG.info("updated columns in row: " + Bytes.toString(newRegion.getRegionName()));
}
}
/** Instantiated to compact the meta region */
private static class OfflineMerger extends Merger {
private final List<HRegionInfo> metaRegions = new ArrayList<HRegionInfo>();
private final HRegion root;
OfflineMerger(Configuration conf, FileSystem fs) throws IOException {
super(conf, fs, HConstants.META_TABLE_NAME);
Path rootTableDir = HTableDescriptor.getTableDir(fs.makeQualified(new Path(conf
.get(HConstants.HBASE_DIR))), HConstants.ROOT_TABLE_NAME);
// Scan root region to find all the meta regions
root = HRegion.newHRegion(rootTableDir, hlog, fs, conf, HRegionInfo.ROOT_REGIONINFO,
null);
root.initialize();
Scan scan = new Scan();
scan.addColumn(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER);
InternalScanner rootScanner = root.getScanner(scan);
try {
List<KeyValue> results = new ArrayList<KeyValue>();
while (rootScanner.next(results)) {
for (KeyValue kv : results) {
HRegionInfo info = Writables.getHRegionInfoOrNull(kv.getValue());
if (info != null) {
metaRegions.add(info);
}
}
}
} finally {
rootScanner.close();
try {
root.close();
} catch (IOException e) {
LOG.error(e);
}
}
}
@Override
protected HRegionInfo[] next() {
HRegionInfo[] results = null;
if (metaRegions.size() > 0) {
results = metaRegions.toArray(new HRegionInfo[metaRegions.size()]);
metaRegions.clear();
}
return results;
}
@Override
protected void updateMeta(final byte[] oldRegion1, final byte[] oldRegion2,
HRegion newRegion) throws IOException {
byte[][] regionsToDelete = { oldRegion1, oldRegion2 };
for (int r = 0; r < regionsToDelete.length; r++) {
Delete delete = new Delete(regionsToDelete[r]);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SERVER_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.STARTCODE_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SPLITA_QUALIFIER);
delete.deleteColumns(HConstants.CATALOG_FAMILY, HConstants.SPLITB_QUALIFIER);
root.delete(delete, null, true);
LOG.info("updated columns in row: " + Bytes.toString(regionsToDelete[r]));
}
HRegionInfo newInfo = newRegion.getRegionInfo();
newInfo.setOffline(true);
Put put = new Put(newRegion.getRegionName());
put.add(HConstants.CATALOG_FAMILY, HConstants.REGIONINFO_QUALIFIER, Writables
.getBytes(newInfo));
root.put(put);
LOG.info("updated columns in row: " + Bytes.toString(newRegion.getRegionName()));
}
}
}
发表评论
-
hbase关于bloom filter使用
2013-08-14 11:36 16971.Bloomfilter的原理? 可参考 http://hi ... -
hbase常用运维命令
2012-03-09 10:21 01.停止运行某个RegionServer,在该RegionSe ... -
HBase bulkload的一个bug定位
2011-12-30 11:13 2300在HBase bulkload过程中,如果rowkey和ver ... -
HBase 不常用命令列表
2011-12-30 10:38 12351.查看hfile的内容 hbase org.apache. ... -
Hbase的Region Compact算法实现分析
2011-12-08 15:06 4665Hbase的Region Compact算法属于一种多路归并的 ... -
Hbase region split源代码阅读笔记
2011-12-07 00:03 3210客户端 1. HbaseAdmin.split(final b ... -
Hbase bulkload源代码阅读笔记
2011-12-06 00:21 60451.LoadIncrementalHFiles.doBulkL ... -
Hbase HLog源代码阅读笔记
2011-11-03 18:01 2868HLog 当客户端往RegionServer上提交了一个更新操 ... -
Hbase WAL原理学习
2011-10-31 20:19 62591.概述 客户端往RegionServer端提交数据的时候,会 ... -
Hbase put过程源代码阅读笔记
2011-10-27 10:22 3454客户端 1.HTable.put for (Put p ... -
Hbase region compact源代码阅读笔记
2011-10-08 19:49 2948边缘代码略过,核心代 ...
相关推荐
5. **故障恢复和高可用性**:通过研究源代码中的故障检测和恢复机制,了解HBase如何保证服务的连续性和数据的一致性。 6. **监控和调优**:源代码中包含了监控和性能调优的相关工具和方法,这对于大型生产环境的...
这本书的源代码包含了丰富的示例和实践案例,对于想要深入了解HBase工作原理和技术细节的开发者来说,无疑是一份宝贵的资源。 首先,我们来了解一下HBase的核心概念。HBase是基于列族的存储模型,这意味着数据被...
通过阅读和理解这段代码,可以深入学习如何在实际项目中将HDFS数据导入HBase,这对于大数据平台的开发和运维人员来说是非常有价值的实践。 总的来说,将HDFS上的文件导入HBase是一个涉及数据处理、数据库设计和编程...
HBase是Apache Hadoop生态系统中的一个分布式、高性能、版本化、列族式数据库,它主要设计用于处理海量数据...同时,了解源代码有助于开发者更好地利用HBase特性,优化数据访问性能,以及在遇到问题时进行排查和解决。
本文将基于hbase-0.98.23的源代码,深入解析其内部机制,帮助读者理解HBase的运行原理和设计思想。 首先,HBase的核心设计理念是基于BigTable模型,它将数据存储为表,每个表由行和列族组成。行键是唯一的,而列族...
VC代码 hbase1.0 (实用代码源).rarVC代码 hbase1.0 (实用代码源).rarVC代码 hbase1.0 (实用代码源).rarVC代码 hbase1.0 (实用代码源).rarVC代码 hbase1.0 (实用代码源).rarVC代码 hbase1.0 (实用代码源).rarVC代码 ...
通过阅读《HBase权威指南》并结合源代码,开发者能够更好地掌握HBase的核心概念和技术,从而在实际项目中有效地运用HBase解决大数据问题。这些源代码实例对于学习HBase的开发、调试和运维都具有很高的参考价值。
基于Hadoop的大数据编程,主要涉及HDFS,MapRedue,HBase+源代码+文档说明 -------- 不懂运行,下载完可以私聊问,可远程教学 该资源内项目源码是个人的毕设,代码都测试ok,都是运行成功后才上传资源,答辩评审平均分...
Hbase权威指南 随书源代码 源码包 绝对完整版 maven工程,带pom文件,可以直接作为一个完整工程导入eclipse等ide。
8. **源码解析**:深入分析WordCount和HBase MapReduce样例代码的细节,帮助读者理解并学习如何自己编写类似的应用。 由于具体的文章内容没有给出,以上都是基于标题和标签的推测。在实际的学习过程中,理解并实践...
通过阅读《HBase 权威指南》的源代码,开发者不仅可以理解HBase的设计理念,还能学习到如何利用HBase解决实际问题,提升大数据处理的能力。无论是初学者还是经验丰富的开发者,这本书都将是提升HBase技能的宝贵资源...
hbase操作必备客户端源代码
在Java中操作HBase数据库时,我们经常需要对大量数据进行高效的检索和处理。...提供的源代码应该包含如何创建、应用PageFilter以及处理扫描结果的完整示例,这对于学习和实践HBase的分页查询非常有帮助。
在HBase这个分布式列式数据库中,Java是主要的编程语言,用于开发和操作HBase。HBase 0.96版本是一个较早但仍然重要的发行版,它引入了许多功能和改进,使得处理大规模数据变得更加高效。下面我们将深入探讨HBase ...
关于压缩包内的子文件,"hbase-2.0.0-alpha3"可能指的是HBase的源代码或者二进制发行版。如果是源代码,开发者可以深入理解HBase的工作原理,甚至对其进行定制或贡献代码。如果是二进制发行版,用户可以直接编译安装...
这个压缩包“hbase-0.94.13.rar”包含了HBase 0.94.13版本的jar包和源代码,对于学习和开发基于HBase的应用程序非常有帮助。 首先,`hbase-0.94.13.jar`是HBase的二进制库文件,包含了编译好的Java类,可以直接在...
解压后的`hbase-book-master`包含项目的基本目录结构,如`src/main/java`用于存放Java源代码,`src/main/resources`存储资源配置文件,`pom.xml`是Maven项目对象模型,定义了项目的构建过程和依赖关系。 2. **...
HBase基本操作 增删改查 java代码 要使用须导入对应的jar包
《HBase权威指南》是一本深入探讨分布式列式数据库HBase的专业书籍,其配套源码提供了书中所提及的示例代码和实践案例,方便读者更好地理解和应用HBase。以下将详细解析HBase的相关知识点。 HBase是建立在Apache ...