Search in sources :

Example 1 with InputSampler

use of org.apache.hadoop.mapreduce.lib.partition.InputSampler in project hadoop by apache.

the class Sort method run.

/**
   * The main driver for sort program.
   * Invoke this method to submit the map/reduce job.
   * @throws IOException When there is communication problems with the 
   *                     job tracker.
   */
public int run(String[] args) throws Exception {
    Configuration conf = getConf();
    JobClient client = new JobClient(conf);
    ClusterStatus cluster = client.getClusterStatus();
    int num_reduces = (int) (cluster.getMaxReduceTasks() * 0.9);
    String sort_reduces = conf.get(REDUCES_PER_HOST);
    if (sort_reduces != null) {
        num_reduces = cluster.getTaskTrackers() * Integer.parseInt(sort_reduces);
    }
    Class<? extends InputFormat> inputFormatClass = SequenceFileInputFormat.class;
    Class<? extends OutputFormat> outputFormatClass = SequenceFileOutputFormat.class;
    Class<? extends WritableComparable> outputKeyClass = BytesWritable.class;
    Class<? extends Writable> outputValueClass = BytesWritable.class;
    List<String> otherArgs = new ArrayList<String>();
    InputSampler.Sampler<K, V> sampler = null;
    for (int i = 0; i < args.length; ++i) {
        try {
            if ("-r".equals(args[i])) {
                num_reduces = Integer.parseInt(args[++i]);
            } else if ("-inFormat".equals(args[i])) {
                inputFormatClass = Class.forName(args[++i]).asSubclass(InputFormat.class);
            } else if ("-outFormat".equals(args[i])) {
                outputFormatClass = Class.forName(args[++i]).asSubclass(OutputFormat.class);
            } else if ("-outKey".equals(args[i])) {
                outputKeyClass = Class.forName(args[++i]).asSubclass(WritableComparable.class);
            } else if ("-outValue".equals(args[i])) {
                outputValueClass = Class.forName(args[++i]).asSubclass(Writable.class);
            } else if ("-totalOrder".equals(args[i])) {
                double pcnt = Double.parseDouble(args[++i]);
                int numSamples = Integer.parseInt(args[++i]);
                int maxSplits = Integer.parseInt(args[++i]);
                if (0 >= maxSplits)
                    maxSplits = Integer.MAX_VALUE;
                sampler = new InputSampler.RandomSampler<K, V>(pcnt, numSamples, maxSplits);
            } else {
                otherArgs.add(args[i]);
            }
        } catch (NumberFormatException except) {
            System.out.println("ERROR: Integer expected instead of " + args[i]);
            return printUsage();
        } catch (ArrayIndexOutOfBoundsException except) {
            System.out.println("ERROR: Required parameter missing from " + args[i - 1]);
            // exits
            return printUsage();
        }
    }
    // Set user-supplied (possibly default) job configs
    job = Job.getInstance(conf);
    job.setJobName("sorter");
    job.setJarByClass(Sort.class);
    job.setMapperClass(Mapper.class);
    job.setReducerClass(Reducer.class);
    job.setNumReduceTasks(num_reduces);
    job.setInputFormatClass(inputFormatClass);
    job.setOutputFormatClass(outputFormatClass);
    job.setOutputKeyClass(outputKeyClass);
    job.setOutputValueClass(outputValueClass);
    // Make sure there are exactly 2 parameters left.
    if (otherArgs.size() != 2) {
        System.out.println("ERROR: Wrong number of parameters: " + otherArgs.size() + " instead of 2.");
        return printUsage();
    }
    FileInputFormat.setInputPaths(job, otherArgs.get(0));
    FileOutputFormat.setOutputPath(job, new Path(otherArgs.get(1)));
    if (sampler != null) {
        System.out.println("Sampling input to effect total-order sort...");
        job.setPartitionerClass(TotalOrderPartitioner.class);
        Path inputDir = FileInputFormat.getInputPaths(job)[0];
        FileSystem fs = inputDir.getFileSystem(conf);
        inputDir = inputDir.makeQualified(fs.getUri(), fs.getWorkingDirectory());
        Path partitionFile = new Path(inputDir, "_sortPartitioning");
        TotalOrderPartitioner.setPartitionFile(conf, partitionFile);
        InputSampler.<K, V>writePartitionFile(job, sampler);
        URI partitionUri = new URI(partitionFile.toString() + "#" + "_sortPartitioning");
        job.addCacheFile(partitionUri);
    }
    System.out.println("Running on " + cluster.getTaskTrackers() + " nodes to sort from " + FileInputFormat.getInputPaths(job)[0] + " into " + FileOutputFormat.getOutputPath(job) + " with " + num_reduces + " reduces.");
    Date startTime = new Date();
    System.out.println("Job started: " + startTime);
    int ret = job.waitForCompletion(true) ? 0 : 1;
    Date end_time = new Date();
    System.out.println("Job ended: " + end_time);
    System.out.println("The job took " + (end_time.getTime() - startTime.getTime()) / 1000 + " seconds.");
    return ret;
}
Also used : SequenceFileOutputFormat(org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat) InputSampler(org.apache.hadoop.mapreduce.lib.partition.InputSampler) Configuration(org.apache.hadoop.conf.Configuration) Writable(org.apache.hadoop.io.Writable) BytesWritable(org.apache.hadoop.io.BytesWritable) JobClient(org.apache.hadoop.mapred.JobClient) URI(java.net.URI) FileSystem(org.apache.hadoop.fs.FileSystem) Path(org.apache.hadoop.fs.Path) SequenceFileInputFormat(org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat) FileOutputFormat(org.apache.hadoop.mapreduce.lib.output.FileOutputFormat) SequenceFileOutputFormat(org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat) BytesWritable(org.apache.hadoop.io.BytesWritable) ClusterStatus(org.apache.hadoop.mapred.ClusterStatus)

Aggregations

URI (java.net.URI)1 Configuration (org.apache.hadoop.conf.Configuration)1 FileSystem (org.apache.hadoop.fs.FileSystem)1 Path (org.apache.hadoop.fs.Path)1 BytesWritable (org.apache.hadoop.io.BytesWritable)1 Writable (org.apache.hadoop.io.Writable)1 ClusterStatus (org.apache.hadoop.mapred.ClusterStatus)1 JobClient (org.apache.hadoop.mapred.JobClient)1 SequenceFileInputFormat (org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat)1 FileOutputFormat (org.apache.hadoop.mapreduce.lib.output.FileOutputFormat)1 SequenceFileOutputFormat (org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat)1 InputSampler (org.apache.hadoop.mapreduce.lib.partition.InputSampler)1