001/** 002 * Licensed to the Apache Software Foundation (ASF) under one 003 * or more contributor license agreements. See the NOTICE file 004 * distributed with this work for additional information 005 * regarding copyright ownership. The ASF licenses this file 006 * to you under the Apache License, Version 2.0 (the 007 * "License"); you may not use this file except in compliance 008 * with the License. You may obtain a copy of the License at 009 * 010 * http://www.apache.org/licenses/LICENSE-2.0 011 * 012 * Unless required by applicable law or agreed to in writing, software 013 * distributed under the License is distributed on an "AS IS" BASIS, 014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 015 * See the License for the specific language governing permissions and 016 * limitations under the License. 017 */ 018 019package org.apache.hadoop.mapred; 020 021import java.io.IOException; 022import java.text.NumberFormat; 023 024import org.apache.hadoop.classification.InterfaceAudience; 025import org.apache.hadoop.classification.InterfaceAudience.Private; 026import org.apache.hadoop.classification.InterfaceStability; 027import org.apache.hadoop.fs.FileSystem; 028import org.apache.hadoop.fs.Path; 029import org.apache.hadoop.io.compress.CompressionCodec; 030import org.apache.hadoop.mapred.FileAlreadyExistsException; 031import org.apache.hadoop.mapreduce.security.TokenCache; 032import org.apache.hadoop.util.Progressable; 033 034/** A base class for {@link OutputFormat}. */ 035@InterfaceAudience.Public 036@InterfaceStability.Stable 037public abstract class FileOutputFormat<K, V> implements OutputFormat<K, V> { 038 039 @Deprecated 040 public static enum Counter { 041 BYTES_WRITTEN 042 } 043 044 /** 045 * Set whether the output of the job is compressed. 046 * @param conf the {@link JobConf} to modify 047 * @param compress should the output of the job be compressed? 048 */ 049 public static void setCompressOutput(JobConf conf, boolean compress) { 050 conf.setBoolean(org.apache.hadoop.mapreduce.lib.output. 051 FileOutputFormat.COMPRESS, compress); 052 } 053 054 /** 055 * Is the job output compressed? 056 * @param conf the {@link JobConf} to look in 057 * @return <code>true</code> if the job output should be compressed, 058 * <code>false</code> otherwise 059 */ 060 public static boolean getCompressOutput(JobConf conf) { 061 return conf.getBoolean(org.apache.hadoop.mapreduce.lib.output. 062 FileOutputFormat.COMPRESS, false); 063 } 064 065 /** 066 * Set the {@link CompressionCodec} to be used to compress job outputs. 067 * @param conf the {@link JobConf} to modify 068 * @param codecClass the {@link CompressionCodec} to be used to 069 * compress the job outputs 070 */ 071 public static void 072 setOutputCompressorClass(JobConf conf, 073 Class<? extends CompressionCodec> codecClass) { 074 setCompressOutput(conf, true); 075 conf.setClass(org.apache.hadoop.mapreduce.lib.output. 076 FileOutputFormat.COMPRESS_CODEC, codecClass, 077 CompressionCodec.class); 078 } 079 080 /** 081 * Get the {@link CompressionCodec} for compressing the job outputs. 082 * @param conf the {@link JobConf} to look in 083 * @param defaultValue the {@link CompressionCodec} to return if not set 084 * @return the {@link CompressionCodec} to be used to compress the 085 * job outputs 086 * @throws IllegalArgumentException if the class was specified, but not found 087 */ 088 public static Class<? extends CompressionCodec> 089 getOutputCompressorClass(JobConf conf, 090 Class<? extends CompressionCodec> defaultValue) { 091 Class<? extends CompressionCodec> codecClass = defaultValue; 092 093 String name = conf.get(org.apache.hadoop.mapreduce.lib.output. 094 FileOutputFormat.COMPRESS_CODEC); 095 if (name != null) { 096 try { 097 codecClass = 098 conf.getClassByName(name).asSubclass(CompressionCodec.class); 099 } catch (ClassNotFoundException e) { 100 throw new IllegalArgumentException("Compression codec " + name + 101 " was not found.", e); 102 } 103 } 104 return codecClass; 105 } 106 107 public abstract RecordWriter<K, V> getRecordWriter(FileSystem ignored, 108 JobConf job, String name, 109 Progressable progress) 110 throws IOException; 111 112 public void checkOutputSpecs(FileSystem ignored, JobConf job) 113 throws FileAlreadyExistsException, 114 InvalidJobConfException, IOException { 115 // Ensure that the output directory is set and not already there 116 Path outDir = getOutputPath(job); 117 if (outDir == null && job.getNumReduceTasks() != 0) { 118 throw new InvalidJobConfException("Output directory not set in JobConf."); 119 } 120 if (outDir != null) { 121 FileSystem fs = outDir.getFileSystem(job); 122 // normalize the output directory 123 outDir = fs.makeQualified(outDir); 124 setOutputPath(job, outDir); 125 126 // get delegation token for the outDir's file system 127 TokenCache.obtainTokensForNamenodes(job.getCredentials(), 128 new Path[] {outDir}, job); 129 130 // check its existence 131 if (fs.exists(outDir)) { 132 throw new FileAlreadyExistsException("Output directory " + outDir + 133 " already exists"); 134 } 135 } 136 } 137 138 /** 139 * Set the {@link Path} of the output directory for the map-reduce job. 140 * 141 * @param conf The configuration of the job. 142 * @param outputDir the {@link Path} of the output directory for 143 * the map-reduce job. 144 */ 145 public static void setOutputPath(JobConf conf, Path outputDir) { 146 outputDir = new Path(conf.getWorkingDirectory(), outputDir); 147 conf.set(org.apache.hadoop.mapreduce.lib.output. 148 FileOutputFormat.OUTDIR, outputDir.toString()); 149 } 150 151 /** 152 * Set the {@link Path} of the task's temporary output directory 153 * for the map-reduce job. 154 * 155 * <p><i>Note</i>: Task output path is set by the framework. 156 * </p> 157 * @param conf The configuration of the job. 158 * @param outputDir the {@link Path} of the output directory 159 * for the map-reduce job. 160 */ 161 @Private 162 public static void setWorkOutputPath(JobConf conf, Path outputDir) { 163 outputDir = new Path(conf.getWorkingDirectory(), outputDir); 164 conf.set(JobContext.TASK_OUTPUT_DIR, outputDir.toString()); 165 } 166 167 /** 168 * Get the {@link Path} to the output directory for the map-reduce job. 169 * 170 * @return the {@link Path} to the output directory for the map-reduce job. 171 * @see FileOutputFormat#getWorkOutputPath(JobConf) 172 */ 173 public static Path getOutputPath(JobConf conf) { 174 String name = conf.get(org.apache.hadoop.mapreduce.lib.output. 175 FileOutputFormat.OUTDIR); 176 return name == null ? null: new Path(name); 177 } 178 179 /** 180 * Get the {@link Path} to the task's temporary output directory 181 * for the map-reduce job 182 * 183 * <h4 id="SideEffectFiles">Tasks' Side-Effect Files</h4> 184 * 185 * <p><i>Note:</i> The following is valid only if the {@link OutputCommitter} 186 * is {@link FileOutputCommitter}. If <code>OutputCommitter</code> is not 187 * a <code>FileOutputCommitter</code>, the task's temporary output 188 * directory is same as {@link #getOutputPath(JobConf)} i.e. 189 * <tt>${mapreduce.output.fileoutputformat.outputdir}$</tt></p> 190 * 191 * <p>Some applications need to create/write-to side-files, which differ from 192 * the actual job-outputs. 193 * 194 * <p>In such cases there could be issues with 2 instances of the same TIP 195 * (running simultaneously e.g. speculative tasks) trying to open/write-to the 196 * same file (path) on HDFS. Hence the application-writer will have to pick 197 * unique names per task-attempt (e.g. using the attemptid, say 198 * <tt>attempt_200709221812_0001_m_000000_0</tt>), not just per TIP.</p> 199 * 200 * <p>To get around this the Map-Reduce framework helps the application-writer 201 * out by maintaining a special 202 * <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_${taskid}</tt> 203 * sub-directory for each task-attempt on HDFS where the output of the 204 * task-attempt goes. On successful completion of the task-attempt the files 205 * in the <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_${taskid}</tt> (only) 206 * are <i>promoted</i> to <tt>${mapreduce.output.fileoutputformat.outputdir}</tt>. Of course, the 207 * framework discards the sub-directory of unsuccessful task-attempts. This 208 * is completely transparent to the application.</p> 209 * 210 * <p>The application-writer can take advantage of this by creating any 211 * side-files required in <tt>${mapreduce.task.output.dir}</tt> during execution 212 * of his reduce-task i.e. via {@link #getWorkOutputPath(JobConf)}, and the 213 * framework will move them out similarly - thus she doesn't have to pick 214 * unique paths per task-attempt.</p> 215 * 216 * <p><i>Note</i>: the value of <tt>${mapreduce.task.output.dir}</tt> during 217 * execution of a particular task-attempt is actually 218 * <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_{$taskid}</tt>, and this value is 219 * set by the map-reduce framework. So, just create any side-files in the 220 * path returned by {@link #getWorkOutputPath(JobConf)} from map/reduce 221 * task to take advantage of this feature.</p> 222 * 223 * <p>The entire discussion holds true for maps of jobs with 224 * reducer=NONE (i.e. 0 reduces) since output of the map, in that case, 225 * goes directly to HDFS.</p> 226 * 227 * @return the {@link Path} to the task's temporary output directory 228 * for the map-reduce job. 229 */ 230 public static Path getWorkOutputPath(JobConf conf) { 231 String name = conf.get(JobContext.TASK_OUTPUT_DIR); 232 return name == null ? null: new Path(name); 233 } 234 235 /** 236 * Helper function to create the task's temporary output directory and 237 * return the path to the task's output file. 238 * 239 * @param conf job-configuration 240 * @param name temporary task-output filename 241 * @return path to the task's temporary output file 242 * @throws IOException 243 */ 244 public static Path getTaskOutputPath(JobConf conf, String name) 245 throws IOException { 246 // ${mapred.out.dir} 247 Path outputPath = getOutputPath(conf); 248 if (outputPath == null) { 249 throw new IOException("Undefined job output-path"); 250 } 251 252 OutputCommitter committer = conf.getOutputCommitter(); 253 Path workPath = outputPath; 254 TaskAttemptContext context = 255 new TaskAttemptContextImpl(conf, 256 TaskAttemptID.forName(conf.get( 257 JobContext.TASK_ATTEMPT_ID))); 258 if (committer instanceof FileOutputCommitter) { 259 workPath = ((FileOutputCommitter)committer).getWorkPath(context, 260 outputPath); 261 } 262 263 // ${mapred.out.dir}/_temporary/_${taskid}/${name} 264 return new Path(workPath, name); 265 } 266 267 /** 268 * Helper function to generate a name that is unique for the task. 269 * 270 * <p>The generated name can be used to create custom files from within the 271 * different tasks for the job, the names for different tasks will not collide 272 * with each other.</p> 273 * 274 * <p>The given name is postfixed with the task type, 'm' for maps, 'r' for 275 * reduces and the task partition number. For example, give a name 'test' 276 * running on the first map o the job the generated name will be 277 * 'test-m-00000'.</p> 278 * 279 * @param conf the configuration for the job. 280 * @param name the name to make unique. 281 * @return a unique name accross all tasks of the job. 282 */ 283 public static String getUniqueName(JobConf conf, String name) { 284 int partition = conf.getInt(JobContext.TASK_PARTITION, -1); 285 if (partition == -1) { 286 throw new IllegalArgumentException( 287 "This method can only be called from within a Job"); 288 } 289 290 String taskType = (conf.getBoolean(JobContext.TASK_ISMAP, true)) ? "m" : "r"; 291 292 NumberFormat numberFormat = NumberFormat.getInstance(); 293 numberFormat.setMinimumIntegerDigits(5); 294 numberFormat.setGroupingUsed(false); 295 296 return name + "-" + taskType + "-" + numberFormat.format(partition); 297 } 298 299 /** 300 * Helper function to generate a {@link Path} for a file that is unique for 301 * the task within the job output directory. 302 * 303 * <p>The path can be used to create custom files from within the map and 304 * reduce tasks. The path name will be unique for each task. The path parent 305 * will be the job output directory.</p>ls 306 * 307 * <p>This method uses the {@link #getUniqueName} method to make the file name 308 * unique for the task.</p> 309 * 310 * @param conf the configuration for the job. 311 * @param name the name for the file. 312 * @return a unique path accross all tasks of the job. 313 */ 314 public static Path getPathForCustomFile(JobConf conf, String name) { 315 return new Path(getWorkOutputPath(conf), getUniqueName(conf, name)); 316 } 317} 318