View Javadoc

1   /**
2    *
3    * Licensed to the Apache Software Foundation (ASF) under one
4    * or more contributor license agreements.  See the NOTICE file
5    * distributed with this work for additional information
6    * regarding copyright ownership.  The ASF licenses this file
7    * to you under the Apache License, Version 2.0 (the
8    * "License"); you may not use this file except in compliance
9    * with the License.  You may obtain a copy of the License at
10   *
11   *     http://www.apache.org/licenses/LICENSE-2.0
12   *
13   * Unless required by applicable law or agreed to in writing, software
14   * distributed under the License is distributed on an "AS IS" BASIS,
15   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16   * See the License for the specific language governing permissions and
17   * limitations under the License.
18   */
19  package org.apache.hadoop.hbase.mapreduce;
20  
21  import java.io.IOException;
22  import java.util.HashMap;
23  import java.util.Map;
24  import java.util.Random;
25  
26  import org.apache.commons.logging.Log;
27  import org.apache.commons.logging.LogFactory;
28  import org.apache.hadoop.conf.Configuration;
29  import org.apache.hadoop.conf.Configured;
30  import org.apache.hadoop.fs.FileSystem;
31  import org.apache.hadoop.fs.Path;
32  import org.apache.hadoop.hbase.HBaseConfiguration;
33  import org.apache.hadoop.hbase.HConstants;
34  import org.apache.hadoop.hbase.TableName;
35  import org.apache.hadoop.hbase.classification.InterfaceAudience;
36  import org.apache.hadoop.hbase.classification.InterfaceStability;
37  import org.apache.hadoop.hbase.client.Admin;
38  import org.apache.hadoop.hbase.client.Connection;
39  import org.apache.hadoop.hbase.client.ConnectionFactory;
40  import org.apache.hadoop.hbase.client.Scan;
41  import org.apache.hadoop.hbase.util.Bytes;
42  import org.apache.hadoop.mapreduce.Job;
43  import org.apache.hadoop.util.GenericOptionsParser;
44  import org.apache.hadoop.util.Tool;
45  import org.apache.hadoop.util.ToolRunner;
46  
47  /**
48   * Tool used to copy a table to another one which can be on a different setup.
49   * It is also configurable with a start and time as well as a specification
50   * of the region server implementation if different from the local cluster.
51   */
52  @InterfaceAudience.Public
53  @InterfaceStability.Stable
54  public class CopyTable extends Configured implements Tool {
55    private static final Log LOG = LogFactory.getLog(CopyTable.class);
56  
57    final static String NAME = "copytable";
58    long startTime = 0;
59    long endTime = 0;
60    int versions = -1;
61    String tableName = null;
62    String startRow = null;
63    String stopRow = null;
64    String dstTableName = null;
65    String peerAddress = null;
66    String families = null;
67    boolean allCells = false;
68    static boolean shuffle = false;
69  
70    boolean bulkload = false;
71    Path bulkloadDir = null;
72  
73    private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";
74  
75    public CopyTable(Configuration conf) {
76      super(conf);
77    }
78    /**
79     * Sets up the actual job.
80     *
81     * @param args  The command line parameters.
82     * @return The newly created job.
83     * @throws IOException When setting up the job fails.
84     */
85    public Job createSubmittableJob(String[] args)
86    throws IOException {
87      if (!doCommandLine(args)) {
88        return null;
89      }
90  
91      Job job = Job.getInstance(getConf(), getConf().get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));
92      job.setJarByClass(CopyTable.class);
93      Scan scan = new Scan();
94      scan.setCacheBlocks(false);
95      if (startTime != 0) {
96        scan.setTimeRange(startTime,
97            endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);
98      }
99      if (allCells) {
100       scan.setRaw(true);
101     }
102     if (shuffle) {
103       job.getConfiguration().set(TableInputFormat.SHUFFLE_MAPS, "true");
104     }
105     if (versions >= 0) {
106       scan.setMaxVersions(versions);
107     }
108 
109     if (startRow != null) {
110       scan.setStartRow(Bytes.toBytes(startRow));
111     }
112 
113     if (stopRow != null) {
114       scan.setStopRow(Bytes.toBytes(stopRow));
115     }
116 
117     if(families != null) {
118       String[] fams = families.split(",");
119       Map<String,String> cfRenameMap = new HashMap<String,String>();
120       for(String fam : fams) {
121         String sourceCf;
122         if(fam.contains(":")) {
123             // fam looks like "sourceCfName:destCfName"
124             String[] srcAndDest = fam.split(":", 2);
125             sourceCf = srcAndDest[0];
126             String destCf = srcAndDest[1];
127             cfRenameMap.put(sourceCf, destCf);
128         } else {
129             // fam is just "sourceCf"
130             sourceCf = fam;
131         }
132         scan.addFamily(Bytes.toBytes(sourceCf));
133       }
134       Import.configureCfRenaming(job.getConfiguration(), cfRenameMap);
135     }
136     job.setNumReduceTasks(0);
137 
138     if (bulkload) {
139       TableMapReduceUtil.initTableMapperJob(tableName, scan, Import.KeyValueImporter.class, null,
140         null, job);
141 
142       // We need to split the inputs by destination tables so that output of Map can be bulk-loaded.
143       TableInputFormat.configureSplitTable(job, TableName.valueOf(dstTableName));
144 
145       FileSystem fs = FileSystem.get(getConf());
146       Random rand = new Random();
147       Path root = new Path(fs.getWorkingDirectory(), "copytable");
148       fs.mkdirs(root);
149       while (true) {
150         bulkloadDir = new Path(root, "" + rand.nextLong());
151         if (!fs.exists(bulkloadDir)) {
152           break;
153         }
154       }
155 
156       System.out.println("HFiles will be stored at " + this.bulkloadDir);
157       HFileOutputFormat2.setOutputPath(job, bulkloadDir);
158       try (Connection conn = ConnectionFactory.createConnection(getConf());
159           Admin admin = conn.getAdmin()) {
160         HFileOutputFormat2.configureIncrementalLoadMap(job,
161             admin.getTableDescriptor((TableName.valueOf(dstTableName))));
162       }
163     } else {
164       TableMapReduceUtil.initTableMapperJob(tableName, scan,
165         Import.Importer.class, null, null, job);
166 
167       TableMapReduceUtil.initTableReducerJob(dstTableName, null, job, null, peerAddress, null,
168         null);
169     }
170 
171     return job;
172   }
173 
174   /*
175    * @param errorMsg Error message.  Can be null.
176    */
177   private static void printUsage(final String errorMsg) {
178     if (errorMsg != null && errorMsg.length() > 0) {
179       System.err.println("ERROR: " + errorMsg);
180     }
181     System.err.println("Usage: CopyTable [general options] [--starttime=X] [--endtime=Y] " +
182         "[--new.name=NEW] [--peer.adr=ADR] <tablename>");
183     System.err.println();
184     System.err.println("Options:");
185     System.err.println(" rs.class     hbase.regionserver.class of the peer cluster");
186     System.err.println("              specify if different from current cluster");
187     System.err.println(" rs.impl      hbase.regionserver.impl of the peer cluster");
188     System.err.println(" startrow     the start row");
189     System.err.println(" stoprow      the stop row");
190     System.err.println(" starttime    beginning of the time range (unixtime in millis)");
191     System.err.println("              without endtime means from starttime to forever");
192     System.err.println(" endtime      end of the time range.  Ignored if no starttime specified.");
193     System.err.println(" versions     number of cell versions to copy");
194     System.err.println(" new.name     new table's name");
195     System.err.println(" peer.adr     Address of the peer cluster given in the format");
196     System.err.println("              hbase.zookeeper.quorum:hbase.zookeeper.client"
197         + ".port:zookeeper.znode.parent");
198     System.err.println(" families     comma-separated list of families to copy");
199     System.err.println("              To copy from cf1 to cf2, give sourceCfName:destCfName. ");
200     System.err.println("              To keep the same name, just give \"cfName\"");
201     System.err.println(" all.cells    also copy delete markers and deleted cells");
202     System.err.println(" bulkload     Write input into HFiles and bulk load to the destination "
203         + "table");
204     System.err.println();
205     System.err.println("Args:");
206     System.err.println(" tablename    Name of the table to copy");
207     System.err.println();
208     System.err.println("Examples:");
209     System.err.println(" To copy 'TestTable' to a cluster that uses replication for a 1 hour window:");
210     System.err.println(" $ bin/hbase " +
211         "org.apache.hadoop.hbase.mapreduce.CopyTable --starttime=1265875194289 --endtime=1265878794289 " +
212         "--peer.adr=server1,server2,server3:2181:/hbase --families=myOldCf:myNewCf,cf2,cf3 TestTable ");
213     System.err.println("For performance consider the following general option:\n"
214         + "  It is recommended that you set the following to >=100. A higher value uses more memory but\n"
215         + "  decreases the round trip time to the server and may increase performance.\n"
216         + "    -Dhbase.client.scanner.caching=100\n"
217         + "  The following should always be set to false, to prevent writing data twice, which may produce \n"
218         + "  inaccurate results.\n"
219         + "    -Dmapreduce.map.speculative=false");
220   }
221 
222   private boolean doCommandLine(final String[] args) {
223     // Process command-line args. TODO: Better cmd-line processing
224     // (but hopefully something not as painful as cli options).
225     if (args.length < 1) {
226       printUsage(null);
227       return false;
228     }
229     try {
230       for (int i = 0; i < args.length; i++) {
231         String cmd = args[i];
232         if (cmd.equals("-h") || cmd.startsWith("--h")) {
233           printUsage(null);
234           return false;
235         }
236 
237         final String startRowArgKey = "--startrow=";
238         if (cmd.startsWith(startRowArgKey)) {
239           startRow = cmd.substring(startRowArgKey.length());
240           continue;
241         }
242 
243         final String stopRowArgKey = "--stoprow=";
244         if (cmd.startsWith(stopRowArgKey)) {
245           stopRow = cmd.substring(stopRowArgKey.length());
246           continue;
247         }
248 
249         final String startTimeArgKey = "--starttime=";
250         if (cmd.startsWith(startTimeArgKey)) {
251           startTime = Long.parseLong(cmd.substring(startTimeArgKey.length()));
252           continue;
253         }
254 
255         final String endTimeArgKey = "--endtime=";
256         if (cmd.startsWith(endTimeArgKey)) {
257           endTime = Long.parseLong(cmd.substring(endTimeArgKey.length()));
258           continue;
259         }
260 
261         final String versionsArgKey = "--versions=";
262         if (cmd.startsWith(versionsArgKey)) {
263           versions = Integer.parseInt(cmd.substring(versionsArgKey.length()));
264           continue;
265         }
266 
267         final String newNameArgKey = "--new.name=";
268         if (cmd.startsWith(newNameArgKey)) {
269           dstTableName = cmd.substring(newNameArgKey.length());
270           continue;
271         }
272 
273         final String peerAdrArgKey = "--peer.adr=";
274         if (cmd.startsWith(peerAdrArgKey)) {
275           peerAddress = cmd.substring(peerAdrArgKey.length());
276           continue;
277         }
278 
279         final String familiesArgKey = "--families=";
280         if (cmd.startsWith(familiesArgKey)) {
281           families = cmd.substring(familiesArgKey.length());
282           continue;
283         }
284 
285         if (cmd.startsWith("--all.cells")) {
286           allCells = true;
287           continue;
288         }
289 
290         if (cmd.startsWith("--bulkload")) {
291           bulkload = true;
292           continue;
293         }
294 
295         if (cmd.startsWith("--shuffle")) {
296           shuffle = true;
297           continue;
298         }
299 
300         if (i == args.length-1) {
301           tableName = cmd;
302         } else {
303           printUsage("Invalid argument '" + cmd + "'");
304           return false;
305         }
306       }
307       if (dstTableName == null && peerAddress == null) {
308         printUsage("At least a new table name or a " +
309             "peer address must be specified");
310         return false;
311       }
312       if ((endTime != 0) && (startTime > endTime)) {
313         printUsage("Invalid time range filter: starttime=" + startTime + " >  endtime=" + endTime);
314         return false;
315       }
316 
317       if (bulkload && peerAddress != null) {
318         printUsage("Remote bulkload is not supported!");
319         return false;
320       }
321 
322       // set dstTableName if necessary
323       if (dstTableName == null) {
324         dstTableName = tableName;
325       }
326     } catch (Exception e) {
327       e.printStackTrace();
328       printUsage("Can't start because " + e.getMessage());
329       return false;
330     }
331     return true;
332   }
333 
334   /**
335    * Main entry point.
336    *
337    * @param args  The command line parameters.
338    * @throws Exception When running the job fails.
339    */
340   public static void main(String[] args) throws Exception {
341     int ret = ToolRunner.run(new CopyTable(HBaseConfiguration.create()), args);
342     System.exit(ret);
343   }
344 
345   @Override
346   public int run(String[] args) throws Exception {
347     String[] otherArgs = new GenericOptionsParser(getConf(), args).getRemainingArgs();
348     Job job = createSubmittableJob(otherArgs);
349     if (job == null) return 1;
350     if (!job.waitForCompletion(true)) {
351       LOG.info("Map-reduce job failed!");
352       if (bulkload) {
353         LOG.info("Files are not bulkloaded!");
354       }
355       return 1;
356     }
357     int code = 0;
358     if (bulkload) {
359       code = new LoadIncrementalHFiles(this.getConf()).run(new String[]{this.bulkloadDir.toString(),
360           this.dstTableName});
361       if (code == 0) {
362         // bulkloadDir is deleted only LoadIncrementalHFiles was successful so that one can rerun
363         // LoadIncrementalHFiles.
364         FileSystem fs = FileSystem.get(this.getConf());
365         if (!fs.delete(this.bulkloadDir, true)) {
366           LOG.error("Deleting folder " + bulkloadDir + " failed!");
367           code = 1;
368         }
369       }
370     }
371     return code;
372   }
373 }