View Javadoc

1   /**
2    * Copyright 2010 The Apache Software Foundation
3    *
4    * Licensed to the Apache Software Foundation (ASF) under one
5    * or more contributor license agreements.  See the NOTICE file
6    * distributed with this work for additional information
7    * regarding copyright ownership.  The ASF licenses this file
8    * to you under the Apache License, Version 2.0 (the
9    * "License"); you may not use this file except in compliance
10   * with the License.  You may obtain a copy of the License at
11   *
12   *     http://www.apache.org/licenses/LICENSE-2.0
13   *
14   * Unless required by applicable law or agreed to in writing, software
15   * distributed under the License is distributed on an "AS IS" BASIS,
16   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
17   * See the License for the specific language governing permissions and
18   * limitations under the License.
19   */
20  package org.apache.hadoop.hbase.mapreduce;
21  
22  import java.io.FileNotFoundException;
23  import java.io.IOException;
24  import java.nio.ByteBuffer;
25  import java.util.ArrayList;
26  import java.util.Arrays;
27  import java.util.Collection;
28  import java.util.Deque;
29  import java.util.HashSet;
30  import java.util.LinkedList;
31  import java.util.List;
32  import java.util.Map;
33  import java.util.Map.Entry;
34  import java.util.Set;
35  import java.util.TreeMap;
36  import java.util.concurrent.Callable;
37  import java.util.concurrent.ExecutionException;
38  import java.util.concurrent.ExecutorService;
39  import java.util.concurrent.Future;
40  import java.util.concurrent.LinkedBlockingQueue;
41  import java.util.concurrent.ThreadPoolExecutor;
42  import java.util.concurrent.TimeUnit;
43  import java.util.concurrent.atomic.AtomicLong;
44  
45  import org.apache.commons.logging.Log;
46  import org.apache.commons.logging.LogFactory;
47  import org.apache.hadoop.conf.Configuration;
48  import org.apache.hadoop.conf.Configured;
49  import org.apache.hadoop.fs.permission.FsPermission;
50  import org.apache.hadoop.fs.FileStatus;
51  import org.apache.hadoop.fs.FileSystem;
52  import org.apache.hadoop.fs.FileUtil;
53  import org.apache.hadoop.fs.Path;
54  import org.apache.hadoop.hbase.HBaseConfiguration;
55  import org.apache.hadoop.hbase.HColumnDescriptor;
56  import org.apache.hadoop.hbase.HConstants;
57  import org.apache.hadoop.hbase.HTableDescriptor;
58  import org.apache.hadoop.hbase.KeyValue;
59  import org.apache.hadoop.hbase.TableNotFoundException;
60  import org.apache.hadoop.hbase.client.HBaseAdmin;
61  import org.apache.hadoop.hbase.client.HConnection;
62  import org.apache.hadoop.hbase.client.HTable;
63  import org.apache.hadoop.hbase.client.ServerCallable;
64  import org.apache.hadoop.hbase.client.UserProvider;
65  import org.apache.hadoop.hbase.coprocessor.SecureBulkLoadClient;
66  import org.apache.hadoop.hbase.io.HalfStoreFileReader;
67  import org.apache.hadoop.hbase.io.Reference;
68  import org.apache.hadoop.hbase.io.Reference.Range;
69  import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;
70  import org.apache.hadoop.hbase.io.hfile.CacheConfig;
71  import org.apache.hadoop.hbase.io.hfile.Compression.Algorithm;
72  import org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoder;
73  import org.apache.hadoop.hbase.io.hfile.HFileDataBlockEncoderImpl;
74  import org.apache.hadoop.hbase.io.hfile.HFile;
75  import org.apache.hadoop.hbase.io.hfile.HFileScanner;
76  import org.apache.hadoop.hbase.regionserver.Store;
77  import org.apache.hadoop.hbase.regionserver.StoreFile;
78  import org.apache.hadoop.hbase.regionserver.StoreFile.BloomType;
79  import org.apache.hadoop.hbase.security.User;
80  import org.apache.hadoop.hbase.util.Bytes;
81  import org.apache.hadoop.hbase.util.Pair;
82  import org.apache.hadoop.security.token.Token;
83  import org.apache.hadoop.util.Tool;
84  import org.apache.hadoop.util.ToolRunner;
85  
86  import com.google.common.collect.HashMultimap;
87  import com.google.common.collect.Multimap;
88  import com.google.common.collect.Multimaps;
89  import com.google.common.util.concurrent.ThreadFactoryBuilder;
90  
91  /**
92   * Tool to load the output of HFileOutputFormat into an existing table.
93   * @see #usage()
94   */
95  public class LoadIncrementalHFiles extends Configured implements Tool {
96  
97    private static Log LOG = LogFactory.getLog(LoadIncrementalHFiles.class);
98    static AtomicLong regionCount = new AtomicLong(0);
99    private HBaseAdmin hbAdmin;
100   private Configuration cfg;
101 
102   public static String NAME = "completebulkload";
103   public static String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";
104 
105   private boolean useSecure;
106   private boolean hasForwardedToken;
107   private Token<?> userToken;
108   private String bulkToken;
109   private final boolean assignSeqIds;
110   private UserProvider userProvider;
111 
112   public LoadIncrementalHFiles(Configuration conf) throws Exception {
113     this(conf, false);
114   }
115 
116   public LoadIncrementalHFiles(Configuration conf, boolean useSecureHBaseOverride) throws Exception {
117     super(conf);
118     this.cfg = conf;
119     this.hbAdmin = new HBaseAdmin(conf);
120     //added simple for testing
121     this.userProvider = UserProvider.instantiate(conf);
122     this.useSecure = useSecureHBaseOverride || userProvider.isHBaseSecurityEnabled();
123     this.assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, false);
124   }
125 
126   private void usage() {
127     System.err.println("usage: " + NAME +
128         " /path/to/hfileoutputformat-output " +
129         "tablename");
130   }
131 
132   /**
133    * Represents an HFile waiting to be loaded. An queue is used
134    * in this class in order to support the case where a region has
135    * split during the process of the load. When this happens,
136    * the HFile is split into two physical parts across the new
137    * region boundary, and each part is added back into the queue.
138    * The import process finishes when the queue is empty.
139    */
140   static class LoadQueueItem {
141     final byte[] family;
142     final Path hfilePath;
143 
144     public LoadQueueItem(byte[] family, Path hfilePath) {
145       this.family = family;
146       this.hfilePath = hfilePath;
147     }
148 
149     public String toString() {
150       return "family:"+ Bytes.toString(family) + " path:" + hfilePath.toString();
151     }
152   }
153 
154   /**
155    * Walk the given directory for all HFiles, and return a Queue
156    * containing all such files.
157    */
158   private void discoverLoadQueue(Deque<LoadQueueItem> ret, Path hfofDir)
159   throws IOException {
160     FileSystem fs = hfofDir.getFileSystem(getConf());
161 
162     if (!fs.exists(hfofDir)) {
163       throw new FileNotFoundException("HFileOutputFormat dir " +
164           hfofDir + " not found");
165     }
166 
167     FileStatus[] familyDirStatuses = fs.listStatus(hfofDir);
168     if (familyDirStatuses == null) {
169       throw new FileNotFoundException("No families found in " + hfofDir);
170     }
171 
172     for (FileStatus stat : familyDirStatuses) {
173       if (!stat.isDir()) {
174         LOG.warn("Skipping non-directory " + stat.getPath());
175         continue;
176       }
177       Path familyDir = stat.getPath();
178       // Skip _logs, etc
179       if (familyDir.getName().startsWith("_")) continue;
180       byte[] family = familyDir.getName().getBytes();
181       Path[] hfiles = FileUtil.stat2Paths(fs.listStatus(familyDir));
182       for (Path hfile : hfiles) {
183         if (hfile.getName().startsWith("_")) continue;
184         ret.add(new LoadQueueItem(family, hfile));
185       }
186     }
187   }
188 
189   /**
190    * Perform a bulk load of the given directory into the given
191    * pre-existing table.  This method is not threadsafe.
192    * 
193    * @param hfofDir the directory that was provided as the output path
194    * of a job using HFileOutputFormat
195    * @param table the table to load into
196    * @throws TableNotFoundException if table does not yet exist
197    */
198   public void doBulkLoad(Path hfofDir, final HTable table)
199     throws TableNotFoundException, IOException
200   {
201     final HConnection conn = table.getConnection();
202 
203     if (!conn.isTableAvailable(table.getTableName())) {
204       throw new TableNotFoundException("Table " +
205           Bytes.toStringBinary(table.getTableName()) +
206           "is not currently available.");
207     }
208 
209     // initialize thread pools
210     int nrThreads = cfg.getInt("hbase.loadincremental.threads.max",
211         Runtime.getRuntime().availableProcessors());
212     ThreadFactoryBuilder builder = new ThreadFactoryBuilder();
213     builder.setNameFormat("LoadIncrementalHFiles-%1$d");
214     ExecutorService pool = new ThreadPoolExecutor(nrThreads, nrThreads,
215         60, TimeUnit.SECONDS,
216         new LinkedBlockingQueue<Runnable>(),
217         builder.build());
218     ((ThreadPoolExecutor)pool).allowCoreThreadTimeOut(true);
219 
220     // LQI queue does not need to be threadsafe -- all operations on this queue
221     // happen in this thread
222     Deque<LoadQueueItem> queue = new LinkedList<LoadQueueItem>();
223     try {
224       discoverLoadQueue(queue, hfofDir);
225       int count = 0;
226 
227       if (queue.isEmpty()) {
228         LOG.warn("Bulk load operation did not find any files to load in " +
229             "directory " + hfofDir.toUri() + ".  Does it contain files in " +
230             "subdirectories that correspond to column family names?");
231         return;
232       }
233 
234       //If using secure bulk load
235       //prepare staging directory and token
236       if(useSecure) {
237         //This condition is here for unit testing
238         //Since delegation token doesn't work in mini cluster
239         if (userProvider.isHadoopSecurityEnabled()) {
240           FileSystem fs = FileSystem.get(cfg);
241           userToken = userProvider.getCurrent().getToken("HDFS_DELEGATION_TOKEN",
242                                                          fs.getCanonicalServiceName());
243           if (userToken == null) {
244             hasForwardedToken = false;
245             userToken = fs.getDelegationToken("renewer");
246           } else {
247             hasForwardedToken = true;
248             LOG.info("Use the existing token: " + userToken);
249           }
250         }
251         bulkToken = new SecureBulkLoadClient(table).prepareBulkLoad(table.getTableName());
252       }
253 
254       // Assumes that region splits can happen while this occurs.
255       while (!queue.isEmpty()) {
256         // need to reload split keys each iteration.
257         final Pair<byte[][], byte[][]> startEndKeys = table.getStartEndKeys();
258         if (count != 0) {
259           LOG.info("Split occured while grouping HFiles, retry attempt " +
260               + count + " with " + queue.size() + " files remaining to group or split");
261         }
262 
263         int maxRetries = cfg.getInt("hbase.bulkload.retries.number", 0);
264         if (maxRetries != 0 && count >= maxRetries) {
265           LOG.error("Retry attempted " + count +  " times without completing, bailing out");
266           return;
267         }
268         count++;
269 
270         // Using ByteBuffer for byte[] equality semantics
271         Multimap<ByteBuffer, LoadQueueItem> regionGroups = groupOrSplitPhase(table,
272             pool, queue, startEndKeys);
273 
274         bulkLoadPhase(table, conn, pool, queue, regionGroups);
275 
276         // NOTE: The next iteration's split / group could happen in parallel to
277         // atomic bulkloads assuming that there are splits and no merges, and
278         // that we can atomically pull out the groups we want to retry.
279       }
280 
281     } finally {
282       if(useSecure) {
283         if(userToken != null && !hasForwardedToken) {
284           try {
285             userToken.cancel(cfg);
286           } catch (Exception e) {
287             LOG.warn("Failed to cancel HDFS delegation token.", e);
288           }
289         }
290         if(bulkToken != null) {
291           new SecureBulkLoadClient(table).cleanupBulkLoad(bulkToken);
292         }
293       }
294       pool.shutdown();
295       if (queue != null && !queue.isEmpty()) {
296         StringBuilder err = new StringBuilder();
297         err.append("-------------------------------------------------\n");
298         err.append("Bulk load aborted with some files not yet loaded:\n");
299         err.append("-------------------------------------------------\n");
300         for (LoadQueueItem q : queue) {
301           err.append("  ").append(q.hfilePath).append('\n');
302         }
303         LOG.error(err);
304       }
305     }
306 
307     if (queue != null && !queue.isEmpty()) {
308         throw new RuntimeException("Bulk load aborted with some files not yet loaded."
309           + "Please check log for more details.");
310     }
311   }
312 
313   /**
314    * This takes the LQI's grouped by likely regions and attempts to bulk load
315    * them.  Any failures are re-queued for another pass with the
316    * groupOrSplitPhase.
317    */
318   protected void bulkLoadPhase(final HTable table, final HConnection conn,
319       ExecutorService pool, Deque<LoadQueueItem> queue,
320       final Multimap<ByteBuffer, LoadQueueItem> regionGroups) throws IOException {
321     // atomically bulk load the groups.
322     Set<Future<List<LoadQueueItem>>> loadingFutures = new HashSet<Future<List<LoadQueueItem>>>();
323     for (Entry<ByteBuffer, ? extends Collection<LoadQueueItem>> e: regionGroups.asMap().entrySet()) {
324       final byte[] first = e.getKey().array();
325       final Collection<LoadQueueItem> lqis =  e.getValue();
326 
327       final Callable<List<LoadQueueItem>> call = new Callable<List<LoadQueueItem>>() {
328         public List<LoadQueueItem> call() throws Exception {
329           List<LoadQueueItem> toRetry = tryAtomicRegionLoad(conn, table.getTableName(), first, lqis);
330           return toRetry;
331         }
332       };
333       loadingFutures.add(pool.submit(call));
334     }
335 
336     // get all the results.
337     for (Future<List<LoadQueueItem>> future : loadingFutures) {
338       try {
339         List<LoadQueueItem> toRetry = future.get();
340 
341         // LQIs that are requeued to be regrouped.
342         queue.addAll(toRetry);
343 
344       } catch (ExecutionException e1) {
345         Throwable t = e1.getCause();
346         if (t instanceof IOException) {
347           // At this point something unrecoverable has happened.
348           // TODO Implement bulk load recovery
349           throw new IOException("BulkLoad encountered an unrecoverable problem", t);
350         }
351         LOG.error("Unexpected execution exception during bulk load", e1);
352         throw new IllegalStateException(t);
353       } catch (InterruptedException e1) {
354         LOG.error("Unexpected interrupted exception during bulk load", e1);
355         throw new IllegalStateException(e1);
356       }
357     }
358   }
359 
360   /**
361    * @return A Multimap<startkey, LoadQueueItem> that groups LQI by likely
362    * bulk load region targets.
363    */
364   private Multimap<ByteBuffer, LoadQueueItem> groupOrSplitPhase(final HTable table,
365       ExecutorService pool, Deque<LoadQueueItem> queue,
366       final Pair<byte[][], byte[][]> startEndKeys) throws IOException {
367     // <region start key, LQI> need synchronized only within this scope of this
368     // phase because of the puts that happen in futures.
369     Multimap<ByteBuffer, LoadQueueItem> rgs = HashMultimap.create();
370     final Multimap<ByteBuffer, LoadQueueItem> regionGroups = Multimaps.synchronizedMultimap(rgs);
371 
372     // drain LQIs and figure out bulk load groups
373     Set<Future<List<LoadQueueItem>>> splittingFutures = new HashSet<Future<List<LoadQueueItem>>>();
374     while (!queue.isEmpty()) {
375       final LoadQueueItem item = queue.remove();
376 
377       final Callable<List<LoadQueueItem>> call = new Callable<List<LoadQueueItem>>() {
378         public List<LoadQueueItem> call() throws Exception {
379           List<LoadQueueItem> splits = groupOrSplit(regionGroups, item, table, startEndKeys);
380           return splits;
381         }
382       };
383       splittingFutures.add(pool.submit(call));
384     }
385     // get all the results.  All grouping and splitting must finish before
386     // we can attempt the atomic loads.
387     for (Future<List<LoadQueueItem>> lqis : splittingFutures) {
388       try {
389         List<LoadQueueItem> splits = lqis.get();
390         if (splits != null) {
391           queue.addAll(splits);
392         }
393       } catch (ExecutionException e1) {
394         Throwable t = e1.getCause();
395         if (t instanceof IOException) {
396           LOG.error("IOException during splitting", e1);
397           throw (IOException)t; // would have been thrown if not parallelized,
398         }
399         LOG.error("Unexpected execution exception during splitting", e1);
400         throw new IllegalStateException(t);
401       } catch (InterruptedException e1) {
402         LOG.error("Unexpected interrupted exception during splitting", e1);
403         throw new IllegalStateException(e1);
404       }
405     }
406     return regionGroups;
407   }
408 
409   // unique file name for the table
410   String getUniqueName(byte[] tableName) {
411     String name = Bytes.toStringBinary(tableName) + "," + regionCount.incrementAndGet();
412     return name;
413   }
414 
415   protected List<LoadQueueItem> splitStoreFile(final LoadQueueItem item,
416       final HTable table, byte[] startKey,
417       byte[] splitKey) throws IOException {
418     final Path hfilePath = item.hfilePath;
419 
420     // We use a '_' prefix which is ignored when walking directory trees
421     // above.
422     final Path tmpDir = new Path(item.hfilePath.getParent(), "_tmp");
423 
424     LOG.info("HFile at " + hfilePath + " no longer fits inside a single " +
425     "region. Splitting...");
426 
427     String uniqueName = getUniqueName(table.getTableName());
428     HColumnDescriptor familyDesc = table.getTableDescriptor().getFamily(item.family);
429     Path botOut = new Path(tmpDir, uniqueName + ".bottom");
430     Path topOut = new Path(tmpDir, uniqueName + ".top");
431     splitStoreFile(getConf(), hfilePath, familyDesc, splitKey,
432         botOut, topOut);
433 
434     FileSystem fs = tmpDir.getFileSystem(getConf());
435     fs.setPermission(tmpDir, FsPermission.valueOf("-rwxrwxrwx"));
436     fs.setPermission(botOut, FsPermission.valueOf("-rwxrwxrwx"));
437 
438     // Add these back at the *front* of the queue, so there's a lower
439     // chance that the region will just split again before we get there.
440     List<LoadQueueItem> lqis = new ArrayList<LoadQueueItem>(2);
441     lqis.add(new LoadQueueItem(item.family, botOut));
442     lqis.add(new LoadQueueItem(item.family, topOut));
443 
444     LOG.info("Successfully split into new HFiles " + botOut + " and " + topOut);
445     return lqis;
446   }
447 
448   /**
449    * Attempt to assign the given load queue item into its target region group.
450    * If the hfile boundary no longer fits into a region, physically splits
451    * the hfile such that the new bottom half will fit and returns the list of
452    * LQI's corresponding to the resultant hfiles.
453    *
454    * protected for testing
455    */
456   protected List<LoadQueueItem> groupOrSplit(Multimap<ByteBuffer, LoadQueueItem> regionGroups,
457       final LoadQueueItem item, final HTable table,
458       final Pair<byte[][], byte[][]> startEndKeys)
459       throws IOException {
460     final Path hfilePath = item.hfilePath;
461     final FileSystem fs = hfilePath.getFileSystem(getConf());
462     HFile.Reader hfr = HFile.createReader(fs, hfilePath,
463         new CacheConfig(getConf()));
464     final byte[] first, last;
465     try {
466       hfr.loadFileInfo();
467       first = hfr.getFirstRowKey();
468       last = hfr.getLastRowKey();
469     }  finally {
470       hfr.close();
471     }
472 
473     LOG.info("Trying to load hfile=" + hfilePath +
474         " first=" + Bytes.toStringBinary(first) +
475         " last="  + Bytes.toStringBinary(last));
476     if (first == null || last == null) {
477       assert first == null && last == null;
478       // TODO what if this is due to a bad HFile?
479       LOG.info("hfile " + hfilePath + " has no entries, skipping");
480       return null;
481     }
482     if (Bytes.compareTo(first, last) > 0) {
483       throw new IllegalArgumentException(
484       "Invalid range: " + Bytes.toStringBinary(first) +
485       " > " + Bytes.toStringBinary(last));
486     }
487     int idx = Arrays.binarySearch(startEndKeys.getFirst(), first,
488         Bytes.BYTES_COMPARATOR);
489     if (idx < 0) {
490       // not on boundary, returns -(insertion index).  Calculate region it
491       // would be in.
492       idx = -(idx + 1) - 1;
493     }
494     final int indexForCallable = idx;
495     
496     /**
497      * we can consider there is a region hole in following conditions. 1) if idx < 0,then first
498      * region info is lost. 2) if the endkey of a region is not equal to the startkey of the next
499      * region. 3) if the endkey of the last region is not empty.
500      */
501     if (indexForCallable < 0) {
502       throw new IOException("The first region info for table "
503           + Bytes.toString(table.getTableName())
504           + " cann't be found in .META..Please use hbck tool to fix it first.");
505     } else if ((indexForCallable == startEndKeys.getFirst().length - 1)
506         && !Bytes.equals(startEndKeys.getSecond()[indexForCallable], HConstants.EMPTY_BYTE_ARRAY)) {
507       throw new IOException("The last region info for table "
508           + Bytes.toString(table.getTableName())
509           + " cann't be found in .META..Please use hbck tool to fix it first.");
510     } else if (indexForCallable + 1 < startEndKeys.getFirst().length
511         && !(Bytes.compareTo(startEndKeys.getSecond()[indexForCallable],
512           startEndKeys.getFirst()[indexForCallable + 1]) == 0)) {
513       throw new IOException("The endkey of one region for table "
514           + Bytes.toString(table.getTableName())
515           + " is not equal to the startkey of the next region in .META.."
516           + "Please use hbck tool to fix it first.");
517     }
518     
519     boolean lastKeyInRange =
520       Bytes.compareTo(last, startEndKeys.getSecond()[idx]) < 0 ||
521       Bytes.equals(startEndKeys.getSecond()[idx], HConstants.EMPTY_BYTE_ARRAY);
522     if (!lastKeyInRange) {
523       List<LoadQueueItem> lqis = splitStoreFile(item, table,
524           startEndKeys.getFirst()[indexForCallable],
525           startEndKeys.getSecond()[indexForCallable]);
526       return lqis;
527     }
528 
529     // group regions.
530     regionGroups.put(ByteBuffer.wrap(startEndKeys.getFirst()[idx]), item);
531     return null;
532   }
533 
534   /**
535    * Attempts to do an atomic load of many hfiles into a region.  If it fails,
536    * it returns a list of hfiles that need to be retried.  If it is successful
537    * it will return an empty list.
538    * 
539    * NOTE: To maintain row atomicity guarantees, region server callable should
540    * succeed atomically and fails atomically.
541    * 
542    * Protected for testing.
543    * 
544    * @return empty list if success, list of items to retry on recoverable
545    * failure
546    */
547   protected List<LoadQueueItem> tryAtomicRegionLoad(final HConnection conn,
548       byte[] tableName, final byte[] first, Collection<LoadQueueItem> lqis) throws IOException {
549 
550     final List<Pair<byte[], String>> famPaths =
551       new ArrayList<Pair<byte[], String>>(lqis.size());
552     for (LoadQueueItem lqi : lqis) {
553       famPaths.add(Pair.newPair(lqi.family, lqi.hfilePath.toString()));
554     }
555 
556     final ServerCallable<Boolean> svrCallable = new ServerCallable<Boolean>(conn,
557         tableName, first) {
558       @Override
559       public Boolean call() throws Exception {
560         SecureBulkLoadClient secureClient = null;
561         boolean success = false;
562 
563         try {
564           LOG.debug("Going to connect to server " + location + " for row "
565               + Bytes.toStringBinary(row));
566           byte[] regionName = location.getRegionInfo().getRegionName();
567           if(!useSecure) {
568              success = server.bulkLoadHFiles(famPaths, regionName, assignSeqIds);
569           } else {
570             HTable table = new HTable(conn.getConfiguration(), tableName);
571             secureClient = new SecureBulkLoadClient(table, location.getRegionInfo().getStartKey());
572             success = secureClient.bulkLoadHFiles(famPaths, userToken, bulkToken, assignSeqIds);
573           }
574           return success;
575         } finally {
576           //Best effort copying of files that might not have been imported
577           //from the staging directory back to original location
578           //in user directory
579           if(secureClient != null && !success) {
580             FileSystem fs = FileSystem.get(cfg);
581             for(Pair<byte[], String> el : famPaths) {
582               Path hfileStagingPath = null;
583               Path hfileOrigPath = new Path(el.getSecond());
584               try {
585                 hfileStagingPath= new Path(secureClient.getStagingPath(bulkToken, el.getFirst()),
586                     hfileOrigPath.getName());
587                 if(fs.rename(hfileStagingPath, hfileOrigPath)) {
588                   LOG.debug("Moved back file " + hfileOrigPath + " from " +
589                       hfileStagingPath);
590                 } else if(fs.exists(hfileStagingPath)){
591                   LOG.debug("Unable to move back file " + hfileOrigPath + " from " +
592                       hfileStagingPath);
593                 }
594               } catch(Exception ex) {
595                 LOG.debug("Unable to move back file " + hfileOrigPath + " from " +
596                     hfileStagingPath, ex);
597               }
598             }
599           }
600         }
601       }
602 
603     };
604 
605     try {
606       List<LoadQueueItem> toRetry = new ArrayList<LoadQueueItem>();
607       boolean success;
608       // secure client wraps the result in another layer of callables, which does its own retrying -
609       // we shouldn't rety again here as well.
610       if (useSecure) {
611         success = svrCallable.withoutRetries();
612       } else {
613         success = svrCallable.withRetries();
614       }
615       if (!success) {
616         LOG.warn("Attempt to bulk load region containing "
617             + Bytes.toStringBinary(first) + " into table "
618             + Bytes.toStringBinary(tableName)  + " with files " + lqis
619             + " failed.  This is recoverable and they will be retried.");
620         toRetry.addAll(lqis); // return lqi's to retry
621       }
622       // success
623       return toRetry;
624     } catch (IOException e) {
625       LOG.error("Encountered unrecoverable error from region server", e);
626       throw e;
627     }
628   }
629 
630   /**
631    * Split a storefile into a top and bottom half, maintaining
632    * the metadata, recreating bloom filters, etc.
633    */
634   static void splitStoreFile(
635       Configuration conf, Path inFile,
636       HColumnDescriptor familyDesc, byte[] splitKey,
637       Path bottomOut, Path topOut) throws IOException
638   {
639     // Open reader with no block cache, and not in-memory
640     Reference topReference = new Reference(splitKey, Range.top);
641     Reference bottomReference = new Reference(splitKey, Range.bottom);
642 
643     copyHFileHalf(conf, inFile, topOut, topReference, familyDesc);
644     copyHFileHalf(conf, inFile, bottomOut, bottomReference, familyDesc);
645   }
646 
647   /**
648    * Copy half of an HFile into a new HFile.
649    */
650   private static void copyHFileHalf(
651       Configuration conf, Path inFile, Path outFile, Reference reference,
652       HColumnDescriptor familyDescriptor)
653   throws IOException {
654     FileSystem fs = inFile.getFileSystem(conf);
655     CacheConfig cacheConf = new CacheConfig(conf);
656     HalfStoreFileReader halfReader = null;
657     StoreFile.Writer halfWriter = null;
658     HFileDataBlockEncoder dataBlockEncoder = new HFileDataBlockEncoderImpl(
659         familyDescriptor.getDataBlockEncodingOnDisk(),
660         familyDescriptor.getDataBlockEncoding());
661     try {
662       halfReader = new HalfStoreFileReader(fs, inFile, cacheConf,
663           reference, DataBlockEncoding.NONE);
664       Map<byte[], byte[]> fileInfo = halfReader.loadFileInfo();
665 
666       int blocksize = familyDescriptor.getBlocksize();
667       Algorithm compression = familyDescriptor.getCompression();
668       BloomType bloomFilterType = familyDescriptor.getBloomFilterType();
669 
670       halfWriter = new StoreFile.WriterBuilder(conf, cacheConf,
671           fs, blocksize)
672               .withFilePath(outFile)
673               .withCompression(compression)
674               .withDataBlockEncoder(dataBlockEncoder)
675               .withBloomType(bloomFilterType)
676               .withChecksumType(Store.getChecksumType(conf))
677               .withBytesPerChecksum(Store.getBytesPerChecksum(conf))
678               .build();
679       HFileScanner scanner = halfReader.getScanner(false, false, false);
680       scanner.seekTo();
681       do {
682         KeyValue kv = scanner.getKeyValue();
683         halfWriter.append(kv);
684       } while (scanner.next());
685 
686       for (Map.Entry<byte[],byte[]> entry : fileInfo.entrySet()) {
687         if (shouldCopyHFileMetaKey(entry.getKey())) {
688           halfWriter.appendFileInfo(entry.getKey(), entry.getValue());
689         }
690       }
691     } finally {
692       if (halfWriter != null) halfWriter.close();
693       if (halfReader != null) halfReader.close(cacheConf.shouldEvictOnClose());
694     }
695   }
696 
697   private static boolean shouldCopyHFileMetaKey(byte[] key) {
698     return !HFile.isReservedFileInfoKey(key);
699   }
700 
701   private boolean doesTableExist(String tableName) throws Exception {
702     return hbAdmin.tableExists(tableName);
703   }
704 
705   /*
706    * Infers region boundaries for a new table.
707    * Parameter:
708    *   bdryMap is a map between keys to an integer belonging to {+1, -1}
709    *     If a key is a start key of a file, then it maps to +1
710    *     If a key is an end key of a file, then it maps to -1
711    * Algo:
712    * 1) Poll on the keys in order: 
713    *    a) Keep adding the mapped values to these keys (runningSum) 
714    *    b) Each time runningSum reaches 0, add the start Key from when the runningSum had started to a boundary list.
715    * 2) Return the boundary list. 
716    */
717   public static byte[][] inferBoundaries(TreeMap<byte[], Integer> bdryMap) {
718     ArrayList<byte[]> keysArray = new ArrayList<byte[]>();
719     int runningValue = 0;
720     byte[] currStartKey = null;
721     boolean firstBoundary = true;
722 
723     for (Map.Entry<byte[], Integer> item: bdryMap.entrySet()) {
724       if (runningValue == 0) currStartKey = item.getKey();
725       runningValue += item.getValue();
726       if (runningValue == 0) {
727         if (!firstBoundary) keysArray.add(currStartKey);
728         firstBoundary = false;
729       }
730     }
731     return keysArray.toArray(new byte[0][0]);
732   }
733  
734   /*
735    * If the table is created for the first time, then "completebulkload" reads the files twice.
736    * More modifications necessary if we want to avoid doing it.
737    */
738   private void createTable(String tableName, String dirPath) throws Exception {
739     Path hfofDir = new Path(dirPath);
740     FileSystem fs = hfofDir.getFileSystem(getConf());
741 
742     if (!fs.exists(hfofDir)) {
743       throw new FileNotFoundException("HFileOutputFormat dir " +
744           hfofDir + " not found");
745     }
746 
747     FileStatus[] familyDirStatuses = fs.listStatus(hfofDir);
748     if (familyDirStatuses == null) {
749       throw new FileNotFoundException("No families found in " + hfofDir);
750     }
751 
752     HTableDescriptor htd = new HTableDescriptor(tableName);
753     HColumnDescriptor hcd = null;
754 
755     // Add column families
756     // Build a set of keys
757     byte[][] keys = null;
758     TreeMap<byte[], Integer> map = new TreeMap<byte[], Integer>(Bytes.BYTES_COMPARATOR);
759 
760     for (FileStatus stat : familyDirStatuses) {
761       if (!stat.isDir()) {
762         LOG.warn("Skipping non-directory " + stat.getPath());
763         continue;
764       }
765       Path familyDir = stat.getPath();
766       // Skip _logs, etc
767       if (familyDir.getName().startsWith("_")) continue;
768       byte[] family = familyDir.getName().getBytes();
769 
770       hcd = new HColumnDescriptor(family);
771       htd.addFamily(hcd);
772 
773       Path[] hfiles = FileUtil.stat2Paths(fs.listStatus(familyDir));
774       for (Path hfile : hfiles) {
775         if (hfile.getName().startsWith("_")) continue;
776         HFile.Reader reader = HFile.createReader(fs, hfile,
777             new CacheConfig(getConf()));
778         final byte[] first, last;
779         try {
780           if (hcd.getCompressionType() != reader.getCompressionAlgorithm()) {
781             hcd.setCompressionType(reader.getCompressionAlgorithm());
782             LOG.info("Setting compression " + hcd.getCompressionType().name() +
783                      " for family " + hcd.toString());
784           }
785           reader.loadFileInfo();
786           first = reader.getFirstRowKey();
787           last =  reader.getLastRowKey();
788 
789           LOG.info("Trying to figure out region boundaries hfile=" + hfile +
790             " first=" + Bytes.toStringBinary(first) +
791             " last="  + Bytes.toStringBinary(last));
792 
793           // To eventually infer start key-end key boundaries
794           Integer value = map.containsKey(first)?(Integer)map.get(first):0;
795           map.put(first, value+1);
796 
797           value = map.containsKey(last)?(Integer)map.get(last):0;
798           map.put(last, value-1);
799         }  finally {
800           reader.close();
801         }
802       }
803     }
804 
805     keys = LoadIncrementalHFiles.inferBoundaries(map);
806     this.hbAdmin.createTable(htd,keys);
807 
808     LOG.info("Table "+ tableName +" is available!!");
809   }
810 
811   @Override
812   public int run(String[] args) throws Exception {
813     if (args.length != 2) {
814       usage();
815       return -1;
816     }
817 
818     String dirPath   = args[0];
819     String tableName = args[1];
820 
821     boolean tableExists   = this.doesTableExist(tableName);
822     if (!tableExists) this.createTable(tableName,dirPath);
823 
824     Path hfofDir = new Path(dirPath);
825     HTable table = new HTable(this.cfg, tableName);
826 
827     doBulkLoad(hfofDir, table);
828     return 0;
829   }
830 
831   public static void main(String[] args) throws Exception {
832     int ret = ToolRunner.run(new LoadIncrementalHFiles(HBaseConfiguration.create()), args);
833     System.exit(ret);
834   }
835 
836 }