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