Search in sources :

Example 6 with ClusterStatus

use of org.apache.hadoop.mapred.ClusterStatus in project whirr by apache.

the class HadoopServiceTest method waitForTaskTrackers.

private static void waitForTaskTrackers(JobClient client) throws IOException {
    while (true) {
        ClusterStatus clusterStatus = client.getClusterStatus();
        int taskTrackerCount = clusterStatus.getTaskTrackers();
        if (taskTrackerCount > 0) {
            break;
        }
        try {
            System.out.print(".");
            Thread.sleep(1000);
        } catch (InterruptedException e) {
            break;
        }
    }
}
Also used : ClusterStatus(org.apache.hadoop.mapred.ClusterStatus)

Example 7 with ClusterStatus

use of org.apache.hadoop.mapred.ClusterStatus in project hadoop-book by elephantscale.

the class RandomWriter method run.

/**
     * This is the main routine for launching a distributed random write job. It
     * runs 10 maps/node and each node writes 1 gig of data to a DFS file. The
     * reduce doesn't do anything.
     *
     * @throws IOException
     */
public int run(String[] args) throws Exception {
    if (args.length == 0) {
        System.out.println("Usage: writer <out-dir>");
        ToolRunner.printGenericCommandUsage(System.out);
        return -1;
    }
    Path outDir = new Path(args[0]);
    JobConf job = new JobConf(getConf());
    job.setJarByClass(RandomWriter.class);
    job.setJobName("random-writer");
    FileOutputFormat.setOutputPath(job, outDir);
    job.setOutputKeyClass(BytesWritable.class);
    job.setOutputValueClass(BytesWritable.class);
    job.setInputFormat(RandomInputFormat.class);
    job.setMapperClass(Map.class);
    job.setReducerClass(IdentityReducer.class);
    job.setOutputFormat(SequenceFileOutputFormat.class);
    JobClient client = new JobClient(job);
    ClusterStatus cluster = client.getClusterStatus();
    int numMapsPerHost = job.getInt("test.randomwriter.maps_per_host", 10);
    long numBytesToWritePerMap = job.getLong("test.randomwrite.bytes_per_map", 1 * 1024 * 1024 * 1024);
    if (numBytesToWritePerMap == 0) {
        System.err.println("Cannot have test.randomwrite.bytes_per_map set to 0");
        return -2;
    }
    long totalBytesToWrite = job.getLong("test.randomwrite.total_bytes", numMapsPerHost * numBytesToWritePerMap * cluster.getTaskTrackers());
    int numMaps = (int) (totalBytesToWrite / numBytesToWritePerMap);
    if (numMaps == 0 && totalBytesToWrite > 0) {
        numMaps = 1;
        job.setLong("test.randomwrite.bytes_per_map", totalBytesToWrite);
    }
    job.setNumMapTasks(numMaps);
    System.out.println("Running " + numMaps + " maps.");
    // reducer NONE
    job.setNumReduceTasks(0);
    Date startTime = new Date();
    System.out.println("Job started: " + startTime);
    JobClient.runJob(job);
    Date endTime = new Date();
    System.out.println("Job ended: " + endTime);
    System.out.println("The job took " + (endTime.getTime() - startTime.getTime()) / 1000 + " seconds.");
    return 0;
}
Also used : Path(org.apache.hadoop.fs.Path) JobConf(org.apache.hadoop.mapred.JobConf) JobClient(org.apache.hadoop.mapred.JobClient) ClusterStatus(org.apache.hadoop.mapred.ClusterStatus) Date(java.util.Date)

Example 8 with ClusterStatus

use of org.apache.hadoop.mapred.ClusterStatus in project hadoop by apache.

the class GenericMRLoadGenerator method confRandom.

/**
   * When no input dir is specified, generate random data.
   */
protected static void confRandom(Job job) throws IOException {
    // from RandomWriter
    job.setInputFormatClass(RandomInputFormat.class);
    job.setMapperClass(RandomMapOutput.class);
    Configuration conf = job.getConfiguration();
    final ClusterStatus cluster = new JobClient(conf).getClusterStatus();
    int numMapsPerHost = conf.getInt(RandomTextWriter.MAPS_PER_HOST, 10);
    long numBytesToWritePerMap = conf.getLong(RandomTextWriter.BYTES_PER_MAP, 1 * 1024 * 1024 * 1024);
    if (numBytesToWritePerMap == 0) {
        throw new IOException("Cannot have " + RandomTextWriter.BYTES_PER_MAP + " set to 0");
    }
    long totalBytesToWrite = conf.getLong(RandomTextWriter.TOTAL_BYTES, numMapsPerHost * numBytesToWritePerMap * cluster.getTaskTrackers());
    int numMaps = (int) (totalBytesToWrite / numBytesToWritePerMap);
    if (numMaps == 0 && totalBytesToWrite > 0) {
        numMaps = 1;
        conf.setLong(RandomTextWriter.BYTES_PER_MAP, totalBytesToWrite);
    }
    conf.setInt(MRJobConfig.NUM_MAPS, numMaps);
}
Also used : Configuration(org.apache.hadoop.conf.Configuration) IOException(java.io.IOException) JobClient(org.apache.hadoop.mapred.JobClient) ClusterStatus(org.apache.hadoop.mapred.ClusterStatus)

Example 9 with ClusterStatus

use of org.apache.hadoop.mapred.ClusterStatus in project hadoop by apache.

the class RandomTextWriter method run.

/**
   * This is the main routine for launching a distributed random write job.
   * It runs 10 maps/node and each node writes 1 gig of data to a DFS file.
   * The reduce doesn't do anything.
   * 
   * @throws IOException 
   */
public int run(String[] args) throws Exception {
    if (args.length == 0) {
        return printUsage();
    }
    Configuration conf = getConf();
    JobClient client = new JobClient(conf);
    ClusterStatus cluster = client.getClusterStatus();
    int numMapsPerHost = conf.getInt(MAPS_PER_HOST, 10);
    long numBytesToWritePerMap = conf.getLong(BYTES_PER_MAP, 1 * 1024 * 1024 * 1024);
    if (numBytesToWritePerMap == 0) {
        System.err.println("Cannot have " + BYTES_PER_MAP + " set to 0");
        return -2;
    }
    long totalBytesToWrite = conf.getLong(TOTAL_BYTES, numMapsPerHost * numBytesToWritePerMap * cluster.getTaskTrackers());
    int numMaps = (int) (totalBytesToWrite / numBytesToWritePerMap);
    if (numMaps == 0 && totalBytesToWrite > 0) {
        numMaps = 1;
        conf.setLong(BYTES_PER_MAP, totalBytesToWrite);
    }
    conf.setInt(MRJobConfig.NUM_MAPS, numMaps);
    Job job = Job.getInstance(conf);
    job.setJarByClass(RandomTextWriter.class);
    job.setJobName("random-text-writer");
    job.setOutputKeyClass(Text.class);
    job.setOutputValueClass(Text.class);
    job.setInputFormatClass(RandomWriter.RandomInputFormat.class);
    job.setMapperClass(RandomTextMapper.class);
    Class<? extends OutputFormat> outputFormatClass = SequenceFileOutputFormat.class;
    List<String> otherArgs = new ArrayList<String>();
    for (int i = 0; i < args.length; ++i) {
        try {
            if ("-outFormat".equals(args[i])) {
                outputFormatClass = Class.forName(args[++i]).asSubclass(OutputFormat.class);
            } else {
                otherArgs.add(args[i]);
            }
        } catch (ArrayIndexOutOfBoundsException except) {
            System.out.println("ERROR: Required parameter missing from " + args[i - 1]);
            // exits
            return printUsage();
        }
    }
    job.setOutputFormatClass(outputFormatClass);
    FileOutputFormat.setOutputPath(job, new Path(otherArgs.get(0)));
    System.out.println("Running " + numMaps + " maps.");
    // reducer NONE
    job.setNumReduceTasks(0);
    Date startTime = new Date();
    System.out.println("Job started: " + startTime);
    int ret = job.waitForCompletion(true) ? 0 : 1;
    Date endTime = new Date();
    System.out.println("Job ended: " + endTime);
    System.out.println("The job took " + (endTime.getTime() - startTime.getTime()) / 1000 + " seconds.");
    return ret;
}
Also used : SequenceFileOutputFormat(org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat) Path(org.apache.hadoop.fs.Path) Configuration(org.apache.hadoop.conf.Configuration) ArrayList(java.util.ArrayList) FileOutputFormat(org.apache.hadoop.mapreduce.lib.output.FileOutputFormat) SequenceFileOutputFormat(org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat) JobClient(org.apache.hadoop.mapred.JobClient) Date(java.util.Date) ClusterStatus(org.apache.hadoop.mapred.ClusterStatus)

Example 10 with ClusterStatus

use of org.apache.hadoop.mapred.ClusterStatus in project hadoop by apache.

the class RandomWriter method run.

/**
   * This is the main routine for launching a distributed random write job.
   * It runs 10 maps/node and each node writes 1 gig of data to a DFS file.
   * The reduce doesn't do anything.
   * 
   * @throws IOException 
   */
public int run(String[] args) throws Exception {
    if (args.length == 0) {
        System.out.println("Usage: writer <out-dir>");
        ToolRunner.printGenericCommandUsage(System.out);
        return 2;
    }
    Path outDir = new Path(args[0]);
    Configuration conf = getConf();
    JobClient client = new JobClient(conf);
    ClusterStatus cluster = client.getClusterStatus();
    int numMapsPerHost = conf.getInt(MAPS_PER_HOST, 10);
    long numBytesToWritePerMap = conf.getLong(BYTES_PER_MAP, 1 * 1024 * 1024 * 1024);
    if (numBytesToWritePerMap == 0) {
        System.err.println("Cannot have" + BYTES_PER_MAP + " set to 0");
        return -2;
    }
    long totalBytesToWrite = conf.getLong(TOTAL_BYTES, numMapsPerHost * numBytesToWritePerMap * cluster.getTaskTrackers());
    int numMaps = (int) (totalBytesToWrite / numBytesToWritePerMap);
    if (numMaps == 0 && totalBytesToWrite > 0) {
        numMaps = 1;
        conf.setLong(BYTES_PER_MAP, totalBytesToWrite);
    }
    conf.setInt(MRJobConfig.NUM_MAPS, numMaps);
    Job job = Job.getInstance(conf);
    job.setJarByClass(RandomWriter.class);
    job.setJobName("random-writer");
    FileOutputFormat.setOutputPath(job, outDir);
    job.setOutputKeyClass(BytesWritable.class);
    job.setOutputValueClass(BytesWritable.class);
    job.setInputFormatClass(RandomInputFormat.class);
    job.setMapperClass(RandomMapper.class);
    job.setReducerClass(Reducer.class);
    job.setOutputFormatClass(SequenceFileOutputFormat.class);
    System.out.println("Running " + numMaps + " maps.");
    // reducer NONE
    job.setNumReduceTasks(0);
    Date startTime = new Date();
    System.out.println("Job started: " + startTime);
    int ret = job.waitForCompletion(true) ? 0 : 1;
    Date endTime = new Date();
    System.out.println("Job ended: " + endTime);
    System.out.println("The job took " + (endTime.getTime() - startTime.getTime()) / 1000 + " seconds.");
    return ret;
}
Also used : Path(org.apache.hadoop.fs.Path) Configuration(org.apache.hadoop.conf.Configuration) JobClient(org.apache.hadoop.mapred.JobClient) ClusterStatus(org.apache.hadoop.mapred.ClusterStatus) Date(java.util.Date)

Aggregations

ClusterStatus (org.apache.hadoop.mapred.ClusterStatus)13 JobClient (org.apache.hadoop.mapred.JobClient)11 Configuration (org.apache.hadoop.conf.Configuration)7 Path (org.apache.hadoop.fs.Path)7 Date (java.util.Date)6 IOException (java.io.IOException)4 JobConf (org.apache.hadoop.mapred.JobConf)4 FileOutputFormat (org.apache.hadoop.mapreduce.lib.output.FileOutputFormat)4 SequenceFileOutputFormat (org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat)4 ArrayList (java.util.ArrayList)3 BytesWritable (org.apache.hadoop.io.BytesWritable)2 Writable (org.apache.hadoop.io.Writable)2 SequenceFileInputFormat (org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat)2 URI (java.net.URI)1 FileSystem (org.apache.hadoop.fs.FileSystem)1 LockException (org.apache.hadoop.hive.ql.lockmgr.LockException)1 AuthorizationException (org.apache.hadoop.hive.ql.metadata.AuthorizationException)1 HiveException (org.apache.hadoop.hive.ql.metadata.HiveException)1 Job (org.apache.hadoop.mapreduce.Job)1 OutputFormat (org.apache.hadoop.mapreduce.OutputFormat)1