`
uestzengting
  • 浏览: 96445 次
  • 性别: Icon_minigender_1
  • 来自: 北京
社区版块
存档分类
最新评论

HBase HMerge源代码阅读和修改

阅读更多
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()));

        }
    }
}
分享到:
评论

相关推荐

    hbase权威指南源代码下载

    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...

    hbase权威指南.源代码

    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...

    将hdfs上的文件导入hbase的源代码

    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...

    hbase0.94java源代码

    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。

    HBase源代码 hbase-0.98.23

    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).rar

    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 ...

    hbase权威指南源代码

    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

    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

    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。

    WordCount,HBase MR样例代码

    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...

    hbase权威指南 源代码 英文

    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

    hbase操作必备客户端源代码

    java操作Hbase之Hbase专用过滤器PageFilter的使用源代码

    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。

    hbase 的java代码 集合 hbase 0.96

    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 ...

    hbase-2.0.0

    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...

    hbase-0.94.13 jar和源码

    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...

    hbase权威指南源码

    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 Java代码

    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 配套源码

    《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 ...

Global site tag (gtag.js) - Google Analytics