/**
* Start the child process to handle the task for us.
*
* @param conf the task's configuration
* @param recordReader the fake record reader to update progress with
* @param output the collector to send output to
* @param reporter the reporter for the task
* @param outputKeyClass the class of the output keys
* @param outputValueClass the class of the output values
* @throws IOException
* @throws InterruptedException
*/
Application(
JobConf conf,
RecordReader<FloatWritable, NullWritable> recordReader,
OutputCollector<K2, V2> output,
Reporter reporter,
Class<? extends K2> outputKeyClass,
Class<? extends V2> outputValueClass)
throws IOException, InterruptedException {
serverSocket = new ServerSocket(0);
Map<String, String> env = new HashMap<String, String>();
// add TMPDIR environment variable with the value of java.io.tmpdir
env.put("TMPDIR", System.getProperty("java.io.tmpdir"));
env.put("hadoop.pipes.command.port", Integer.toString(serverSocket.getLocalPort()));
List<String> cmd = new ArrayList<String>();
String executable = DistributedCache.getLocalCacheFiles(conf)[0].toString();
FileUtil.chmod(executable, "a+x");
cmd.add(executable);
// wrap the command in a stdout/stderr capture
TaskAttemptID taskid = TaskAttemptID.forName(conf.get("mapred.task.id"));
File stdout = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDOUT);
File stderr = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDERR);
long logLength = TaskLog.getTaskLogLength(conf);
cmd = TaskLog.captureOutAndError(cmd, stdout, stderr, logLength);
process = runClient(cmd, env);
clientSocket = serverSocket.accept();
handler = new OutputHandler<K2, V2>(output, reporter, recordReader);
K2 outputKey = (K2) ReflectionUtils.newInstance(outputKeyClass, conf);
V2 outputValue = (V2) ReflectionUtils.newInstance(outputValueClass, conf);
downlink =
new BinaryProtocol<K1, V1, K2, V2>(clientSocket, handler, outputKey, outputValue, conf);
downlink.start();
downlink.setJobConf(conf);
}
