/**
* 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 org.apache.hadoop.mapred.gridmix;
import java.io.IOException;
import java.security.PrivilegedExceptionAction;
import java.util.ArrayList;
import java.util.List;
import java.util.Random;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.io.BytesWritable;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.SequenceFile;
import org.apache.hadoop.mapred.ClusterStatus;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import org.apache.hadoop.mapred.JobClient;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapreduce.lib.input.SequenceFileRecordReader;
import org.apache.hadoop.mapreduce.InputFormat;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.JobContext;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.RecordReader;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;
import org.apache.hadoop.mapreduce.server.tasktracker.TTConfig;
import org.apache.hadoop.security.UserGroupInformation;
/**
* GridmixJob that generates distributed cache files.
* {@link GenerateDistCacheData} expects a list of distributed cache files to be
* generated as input. This list is expected to be stored as a sequence file
* and the filename is expected to be configured using
* {@code gridmix.distcache.file.list}.
* This input file contains the list of distributed cache files and their sizes.
* For each record (i.e. file size and file path) in this input file,
* a file with the specific file size at the specific path is created.
*/
@InterfaceAudience.Private
@InterfaceStability.Evolving
class GenerateDistCacheData extends GridmixJob {
/**
* Number of distributed cache files to be created by gridmix
*/
static final String GRIDMIX_DISTCACHE_FILE_COUNT =
"gridmix.distcache.file.count";
/**
* Total number of bytes to be written to the distributed cache files by
* gridmix. i.e. Sum of sizes of all unique distributed cache files to be
* created by gridmix.
*/
static final String GRIDMIX_DISTCACHE_BYTE_COUNT =
"gridmix.distcache.byte.count";
/**
* The special file created(and used) by gridmix, that contains the list of
* unique distributed cache files that are to be created and their sizes.
*/
static final String GRIDMIX_DISTCACHE_FILE_LIST =
"gridmix.distcache.file.list";
static final String JOB_NAME = "GRIDMIX_GENERATE_DISTCACHE_DATA";
public GenerateDistCacheData(Configuration conf) throws IOException {
super(conf, 0L, JOB_NAME);
}
@Override
public Job call() throws IOException, InterruptedException,
ClassNotFoundException {
UserGroupInformation ugi = UserGroupInformation.getLoginUser();
ugi.doAs( new PrivilegedExceptionAction <Job>() {
public Job run() throws IOException, ClassNotFoundException,
InterruptedException {
job.setMapperClass(GenDCDataMapper.class);
job.setNumReduceTasks(0);
job.setMapOutputKeyClass(NullWritable.class);
job.setMapOutputValueClass(BytesWritable.class);
job.setInputFormatClass(GenDCDataFormat.class);
job.setOutputFormatClass(NullOutputFormat.class);
job.setJarByClass(GenerateDistCacheData.class);
try {
FileInputFormat.addInputPath(job, new Path("ignored"));
} catch (IOException e) {
LOG.error("Error while adding input path ", e);
}
job.submit();
return job;
}
});
return job;
}
@Override
protected boolean canEmulateCompression() {
return false;
}
public static class GenDCDataMapper
extends Mapper<LongWritable, BytesWritable, NullWritable, BytesWritable> {
private BytesWritable val;
private final Random r = new Random();
private FileSystem fs;
@Override
protected void setup(Context context)
throws IOException, InterruptedException {
val = new BytesWritable(new byte[context.getConfiguration().getInt(
GenerateData.GRIDMIX_VAL_BYTES, 1024 * 1024)]);
fs = FileSystem.get(context.getConfiguration());
}
// Create one distributed cache file with the needed file size.
// key is distributed cache file size and
// value is distributed cache file path.
@Override
public void map(LongWritable key, BytesWritable value, Context context)
throws IOException, InterruptedException {
String fileName = new String(value.getBytes(), 0, value.getLength());
Path path = new Path(fileName);
/**
* Create distributed cache file with the permissions 0755.
* Since the private distributed cache directory doesn't have execute
* permission for others, it is OK to set read permission for others for
* the files under that directory and still they will become 'private'
* distributed cache files on the simulated cluster.
*/
FSDataOutputStream dos =
FileSystem.create(fs, path, new FsPermission((short)0755));
for (long bytes = key.get(); bytes > 0; bytes -= val.getLength()) {
r.nextBytes(val.getBytes());
val.setSize((int)Math.min(val.getLength(), bytes));
dos.write(val.getBytes(), 0, val.getLength());// Write to distCache file
}
dos.close();
}
}
/**
* InputFormat for GenerateDistCacheData.
* Input to GenerateDistCacheData is the special file(in SequenceFile format)
* that contains the list of distributed cache files to be generated along
* with their file sizes.
*/
static class GenDCDataFormat
extends InputFormat<LongWritable, BytesWritable> {
// Split the special file that contains the list of distributed cache file
// paths and their file sizes such that each split corresponds to
// approximately same amount of distributed cache data to be generated.
// Consider numTaskTrackers * numMapSlotsPerTracker as the number of maps
// for this job, if there is lot of data to be generated.
@Override
public List<InputSplit> getSplits(JobContext jobCtxt) throws IOException {
final JobConf jobConf = new JobConf(jobCtxt.getConfiguration());
final JobClient client = new JobClient(jobConf);
ClusterStatus stat = client.getClusterStatus(true);
int numTrackers = stat.getTaskTrackers();
final int fileCount = jobConf.getInt(GRIDMIX_DISTCACHE_FILE_COUNT, -1);
// Total size of distributed cache files to be generated
final long totalSize = jobConf.getLong(GRIDMIX_DISTCACHE_BYTE_COUNT, -1);
// Get the path of the special file
String distCacheFileList = jobConf.get(GRIDMIX_DISTCACHE_FILE_LIST);
if (fileCount < 0 || totalSize < 0 || distCacheFileList == null) {
throw new RuntimeException("Invalid metadata: #files (" + fileCount
+ "), total_size (" + totalSize + "), filelisturi ("
+ distCacheFileList + ")");
}
Path sequenceFile = new Path(distCacheFileList);
FileSystem fs = sequenceFile.getFileSystem(jobConf);
FileStatus srcst = fs.getFileStatus(sequenceFile);
// Consider the number of TTs * mapSlotsPerTracker as number of mappers.
int numMapSlotsPerTracker = jobConf.getInt(TTConfig.TT_MAP_SLOTS, 2);
int numSplits = numTrackers * numMapSlotsPerTracker;
List<InputSplit> splits = new ArrayList<InputSplit>(numSplits);
LongWritable key = new LongWritable();
BytesWritable value = new BytesWritable();
// Average size of data to be generated by each map task
final long targetSize = Math.max(totalSize / numSplits,
DistributedCacheEmulator.AVG_BYTES_PER_MAP);
long splitStartPosition = 0L;
long splitEndPosition = 0L;
long acc = 0L;
long bytesRemaining = srcst.getLen();
SequenceFile.Reader reader = null;
try {
reader = new SequenceFile.Reader(fs, sequenceFile, jobConf);
while (reader.next(key, value)) {
// If adding this file would put this split past the target size,
// cut the last split and put this file in the next split.
if (acc + key.get() > targetSize && acc != 0) {
long splitSize = splitEndPosition - splitStartPosition;
splits.add(new FileSplit(
sequenceFile, splitStartPosition, splitSize, (String[])null));
bytesRemaining -= splitSize;
splitStartPosition = splitEndPosition;
acc = 0L;
}
acc += key.get();
splitEndPosition = reader.getPosition();
}
} finally {
if (reader != null) {
reader.close();
}
}
if (bytesRemaining != 0) {
splits.add(new FileSplit(
sequenceFile, splitStartPosition, bytesRemaining, (String[])null));
}
return splits;
}
/**
* Returns a reader for this split of the distributed cache file list.
*/
@Override
public RecordReader<LongWritable, BytesWritable> createRecordReader(
InputSplit split, final TaskAttemptContext taskContext)
throws IOException, InterruptedException {
return new SequenceFileRecordReader<LongWritable, BytesWritable>();
}
}
}