Examples with DataOutputBuffer org.apache.hadoop.io.DataOutputBuffer used on opensource projects

Example 6 with DataOutputBuffer

use of org.apache.hadoop.io.DataOutputBuffer in project hadoop by apache.

the class TestMRSequenceFileAsBinaryOutputFormat method testBinary.

@Test
public void testBinary() throws IOException, InterruptedException {
    Configuration conf = new Configuration();
    Job job = Job.getInstance(conf);
    Path outdir = new Path(System.getProperty("test.build.data", "/tmp"), "outseq");
    Random r = new Random();
    long seed = r.nextLong();
    r.setSeed(seed);
    FileOutputFormat.setOutputPath(job, outdir);
    SequenceFileAsBinaryOutputFormat.setSequenceFileOutputKeyClass(job, IntWritable.class);
    SequenceFileAsBinaryOutputFormat.setSequenceFileOutputValueClass(job, DoubleWritable.class);
    SequenceFileAsBinaryOutputFormat.setCompressOutput(job, true);
    SequenceFileAsBinaryOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);
    BytesWritable bkey = new BytesWritable();
    BytesWritable bval = new BytesWritable();
    TaskAttemptContext context = MapReduceTestUtil.createDummyMapTaskAttemptContext(job.getConfiguration());
    OutputFormat<BytesWritable, BytesWritable> outputFormat = new SequenceFileAsBinaryOutputFormat();
    OutputCommitter committer = outputFormat.getOutputCommitter(context);
    committer.setupJob(job);
    RecordWriter<BytesWritable, BytesWritable> writer = outputFormat.getRecordWriter(context);
    IntWritable iwritable = new IntWritable();
    DoubleWritable dwritable = new DoubleWritable();
    DataOutputBuffer outbuf = new DataOutputBuffer();
    LOG.info("Creating data by SequenceFileAsBinaryOutputFormat");
    try {
        for (int i = 0; i < RECORDS; ++i) {
            iwritable = new IntWritable(r.nextInt());
            iwritable.write(outbuf);
            bkey.set(outbuf.getData(), 0, outbuf.getLength());
            outbuf.reset();
            dwritable = new DoubleWritable(r.nextDouble());
            dwritable.write(outbuf);
            bval.set(outbuf.getData(), 0, outbuf.getLength());
            outbuf.reset();
            writer.write(bkey, bval);
        }
    } finally {
        writer.close(context);
    }
    committer.commitTask(context);
    committer.commitJob(job);
    InputFormat<IntWritable, DoubleWritable> iformat = new SequenceFileInputFormat<IntWritable, DoubleWritable>();
    int count = 0;
    r.setSeed(seed);
    SequenceFileInputFormat.setInputPaths(job, outdir);
    LOG.info("Reading data by SequenceFileInputFormat");
    for (InputSplit split : iformat.getSplits(job)) {
        RecordReader<IntWritable, DoubleWritable> reader = iformat.createRecordReader(split, context);
        MapContext<IntWritable, DoubleWritable, BytesWritable, BytesWritable> mcontext = new MapContextImpl<IntWritable, DoubleWritable, BytesWritable, BytesWritable>(job.getConfiguration(), context.getTaskAttemptID(), reader, null, null, MapReduceTestUtil.createDummyReporter(), split);
        reader.initialize(split, mcontext);
        try {
            int sourceInt;
            double sourceDouble;
            while (reader.nextKeyValue()) {
                sourceInt = r.nextInt();
                sourceDouble = r.nextDouble();
                iwritable = reader.getCurrentKey();
                dwritable = reader.getCurrentValue();
                assertEquals("Keys don't match: " + "*" + iwritable.get() + ":" + sourceInt + "*", sourceInt, iwritable.get());
                assertTrue("Vals don't match: " + "*" + dwritable.get() + ":" + sourceDouble + "*", Double.compare(dwritable.get(), sourceDouble) == 0);
                ++count;
            }
        } finally {
            reader.close();
        }
    }
    assertEquals("Some records not found", RECORDS, count);
}

Example 7 with DataOutputBuffer

use of org.apache.hadoop.io.DataOutputBuffer in project hadoop by apache.

the class TestShuffleHandler method testRecoveryFromOtherVersions.

@Test
public void testRecoveryFromOtherVersions() throws IOException {
    final String user = "someuser";
    final ApplicationId appId = ApplicationId.newInstance(12345, 1);
    final File tmpDir = new File(System.getProperty("test.build.data", System.getProperty("java.io.tmpdir")), TestShuffleHandler.class.getName());
    Configuration conf = new Configuration();
    conf.setInt(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY, 0);
    conf.setInt(ShuffleHandler.MAX_SHUFFLE_CONNECTIONS, 3);
    ShuffleHandler shuffle = new ShuffleHandler();
    // emulate aux services startup with recovery enabled
    shuffle.setRecoveryPath(new Path(tmpDir.toString()));
    tmpDir.mkdirs();
    try {
        shuffle.init(conf);
        shuffle.start();
        // setup a shuffle token for an application
        DataOutputBuffer outputBuffer = new DataOutputBuffer();
        outputBuffer.reset();
        Token<JobTokenIdentifier> jt = new Token<JobTokenIdentifier>("identifier".getBytes(), "password".getBytes(), new Text(user), new Text("shuffleService"));
        jt.write(outputBuffer);
        shuffle.initializeApplication(new ApplicationInitializationContext(user, appId, ByteBuffer.wrap(outputBuffer.getData(), 0, outputBuffer.getLength())));
        // verify we are authorized to shuffle
        int rc = getShuffleResponseCode(shuffle, jt);
        Assert.assertEquals(HttpURLConnection.HTTP_OK, rc);
        // emulate shuffle handler restart
        shuffle.close();
        shuffle = new ShuffleHandler();
        shuffle.setRecoveryPath(new Path(tmpDir.toString()));
        shuffle.init(conf);
        shuffle.start();
        // verify we are still authorized to shuffle to the old application
        rc = getShuffleResponseCode(shuffle, jt);
        Assert.assertEquals(HttpURLConnection.HTTP_OK, rc);
        Version version = Version.newInstance(1, 0);
        Assert.assertEquals(version, shuffle.getCurrentVersion());
        // emulate shuffle handler restart with compatible version
        Version version11 = Version.newInstance(1, 1);
        // update version info before close shuffle
        shuffle.storeVersion(version11);
        Assert.assertEquals(version11, shuffle.loadVersion());
        shuffle.close();
        shuffle = new ShuffleHandler();
        shuffle.setRecoveryPath(new Path(tmpDir.toString()));
        shuffle.init(conf);
        shuffle.start();
        // shuffle version will be override by CURRENT_VERSION_INFO after restart
        // successfully.
        Assert.assertEquals(version, shuffle.loadVersion());
        // verify we are still authorized to shuffle to the old application
        rc = getShuffleResponseCode(shuffle, jt);
        Assert.assertEquals(HttpURLConnection.HTTP_OK, rc);
        // emulate shuffle handler restart with incompatible version
        Version version21 = Version.newInstance(2, 1);
        shuffle.storeVersion(version21);
        Assert.assertEquals(version21, shuffle.loadVersion());
        shuffle.close();
        shuffle = new ShuffleHandler();
        shuffle.setRecoveryPath(new Path(tmpDir.toString()));
        shuffle.init(conf);
        try {
            shuffle.start();
            Assert.fail("Incompatible version, should expect fail here.");
        } catch (ServiceStateException e) {
            Assert.assertTrue("Exception message mismatch", e.getMessage().contains("Incompatible version for state DB schema:"));
        }
    } finally {
        if (shuffle != null) {
            shuffle.close();
        }
        FileUtil.fullyDelete(tmpDir);
    }
}

Example 8 with DataOutputBuffer

use of org.apache.hadoop.io.DataOutputBuffer in project hadoop by apache.

the class TestShuffleHandler method testKeepAlive.

@Test(timeout = 10000)
public void testKeepAlive() throws Exception {
    final ArrayList<Throwable> failures = new ArrayList<Throwable>(1);
    Configuration conf = new Configuration();
    conf.setInt(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY, 0);
    conf.setBoolean(ShuffleHandler.SHUFFLE_CONNECTION_KEEP_ALIVE_ENABLED, true);
    // try setting to -ve keep alive timeout.
    conf.setInt(ShuffleHandler.SHUFFLE_CONNECTION_KEEP_ALIVE_TIME_OUT, -100);
    ShuffleHandler shuffleHandler = new ShuffleHandler() {

        @Override
        protected Shuffle getShuffle(final Configuration conf) {
            // replace the shuffle handler with one stubbed for testing
            return new Shuffle(conf) {

                @Override
                protected MapOutputInfo getMapOutputInfo(String mapId, int reduce, String jobId, String user) throws IOException {
                    return null;
                }

                @Override
                protected void verifyRequest(String appid, ChannelHandlerContext ctx, HttpRequest request, HttpResponse response, URL requestUri) throws IOException {
                }

                @Override
                protected void populateHeaders(List<String> mapIds, String jobId, String user, int reduce, HttpRequest request, HttpResponse response, boolean keepAliveParam, Map<String, MapOutputInfo> infoMap) throws IOException {
                    // Send some dummy data (populate content length details)
                    ShuffleHeader header = new ShuffleHeader("attempt_12345_1_m_1_0", 5678, 5678, 1);
                    DataOutputBuffer dob = new DataOutputBuffer();
                    header.write(dob);
                    dob = new DataOutputBuffer();
                    for (int i = 0; i < 100000; ++i) {
                        header.write(dob);
                    }
                    long contentLength = dob.getLength();
                    // disable connectinKeepAliveEnabled if keepAliveParam is available
                    if (keepAliveParam) {
                        connectionKeepAliveEnabled = false;
                    }
                    super.setResponseHeaders(response, keepAliveParam, contentLength);
                }

                @Override
                protected ChannelFuture sendMapOutput(ChannelHandlerContext ctx, Channel ch, String user, String mapId, int reduce, MapOutputInfo info) throws IOException {
                    HttpResponse response = new DefaultHttpResponse(HTTP_1_1, OK);
                    // send a shuffle header and a lot of data down the channel
                    // to trigger a broken pipe
                    ShuffleHeader header = new ShuffleHeader("attempt_12345_1_m_1_0", 5678, 5678, 1);
                    DataOutputBuffer dob = new DataOutputBuffer();
                    header.write(dob);
                    ch.write(wrappedBuffer(dob.getData(), 0, dob.getLength()));
                    dob = new DataOutputBuffer();
                    for (int i = 0; i < 100000; ++i) {
                        header.write(dob);
                    }
                    return ch.write(wrappedBuffer(dob.getData(), 0, dob.getLength()));
                }

                @Override
                protected void sendError(ChannelHandlerContext ctx, HttpResponseStatus status) {
                    if (failures.size() == 0) {
                        failures.add(new Error());
                        ctx.getChannel().close();
                    }
                }

                @Override
                protected void sendError(ChannelHandlerContext ctx, String message, HttpResponseStatus status) {
                    if (failures.size() == 0) {
                        failures.add(new Error());
                        ctx.getChannel().close();
                    }
                }
            };
        }
    };
    shuffleHandler.init(conf);
    shuffleHandler.start();
    String shuffleBaseURL = "http://127.0.0.1:" + shuffleHandler.getConfig().get(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY);
    URL url = new URL(shuffleBaseURL + "/mapOutput?job=job_12345_1&reduce=1&" + "map=attempt_12345_1_m_1_0");
    HttpURLConnection conn = (HttpURLConnection) url.openConnection();
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_NAME, ShuffleHeader.DEFAULT_HTTP_HEADER_NAME);
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_VERSION, ShuffleHeader.DEFAULT_HTTP_HEADER_VERSION);
    conn.connect();
    DataInputStream input = new DataInputStream(conn.getInputStream());
    Assert.assertEquals(HttpHeader.KEEP_ALIVE.asString(), conn.getHeaderField(HttpHeader.CONNECTION.asString()));
    Assert.assertEquals("timeout=1", conn.getHeaderField(HttpHeader.KEEP_ALIVE.asString()));
    Assert.assertEquals(HttpURLConnection.HTTP_OK, conn.getResponseCode());
    ShuffleHeader header = new ShuffleHeader();
    header.readFields(input);
    input.close();
    // For keepAlive via URL
    url = new URL(shuffleBaseURL + "/mapOutput?job=job_12345_1&reduce=1&" + "map=attempt_12345_1_m_1_0&keepAlive=true");
    conn = (HttpURLConnection) url.openConnection();
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_NAME, ShuffleHeader.DEFAULT_HTTP_HEADER_NAME);
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_VERSION, ShuffleHeader.DEFAULT_HTTP_HEADER_VERSION);
    conn.connect();
    input = new DataInputStream(conn.getInputStream());
    Assert.assertEquals(HttpHeader.KEEP_ALIVE.asString(), conn.getHeaderField(HttpHeader.CONNECTION.asString()));
    Assert.assertEquals("timeout=1", conn.getHeaderField(HttpHeader.KEEP_ALIVE.asString()));
    Assert.assertEquals(HttpURLConnection.HTTP_OK, conn.getResponseCode());
    header = new ShuffleHeader();
    header.readFields(input);
    input.close();
}

Example 9 with DataOutputBuffer

use of org.apache.hadoop.io.DataOutputBuffer in project hadoop by apache.

the class TestShuffleHandler method testGetMapOutputInfo.

@Test(timeout = 100000)
public void testGetMapOutputInfo() throws Exception {
    final ArrayList<Throwable> failures = new ArrayList<Throwable>(1);
    Configuration conf = new Configuration();
    conf.setInt(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY, 0);
    conf.setInt(ShuffleHandler.MAX_SHUFFLE_CONNECTIONS, 3);
    conf.set(CommonConfigurationKeysPublic.HADOOP_SECURITY_AUTHENTICATION, "simple");
    UserGroupInformation.setConfiguration(conf);
    File absLogDir = new File("target", TestShuffleHandler.class.getSimpleName() + "LocDir").getAbsoluteFile();
    conf.set(YarnConfiguration.NM_LOCAL_DIRS, absLogDir.getAbsolutePath());
    ApplicationId appId = ApplicationId.newInstance(12345, 1);
    String appAttemptId = "attempt_12345_1_m_1_0";
    String user = "randomUser";
    String reducerId = "0";
    List<File> fileMap = new ArrayList<File>();
    createShuffleHandlerFiles(absLogDir, user, appId.toString(), appAttemptId, conf, fileMap);
    ShuffleHandler shuffleHandler = new ShuffleHandler() {

        @Override
        protected Shuffle getShuffle(Configuration conf) {
            // replace the shuffle handler with one stubbed for testing
            return new Shuffle(conf) {

                @Override
                protected void populateHeaders(List<String> mapIds, String outputBaseStr, String user, int reduce, HttpRequest request, HttpResponse response, boolean keepAliveParam, Map<String, MapOutputInfo> infoMap) throws IOException {
                    // Only set response headers and skip everything else
                    // send some dummy value for content-length
                    super.setResponseHeaders(response, keepAliveParam, 100);
                }

                @Override
                protected void verifyRequest(String appid, ChannelHandlerContext ctx, HttpRequest request, HttpResponse response, URL requestUri) throws IOException {
                // Do nothing.
                }

                @Override
                protected void sendError(ChannelHandlerContext ctx, String message, HttpResponseStatus status) {
                    if (failures.size() == 0) {
                        failures.add(new Error(message));
                        ctx.getChannel().close();
                    }
                }

                @Override
                protected ChannelFuture sendMapOutput(ChannelHandlerContext ctx, Channel ch, String user, String mapId, int reduce, MapOutputInfo info) throws IOException {
                    // send a shuffle header
                    ShuffleHeader header = new ShuffleHeader("attempt_12345_1_m_1_0", 5678, 5678, 1);
                    DataOutputBuffer dob = new DataOutputBuffer();
                    header.write(dob);
                    return ch.write(wrappedBuffer(dob.getData(), 0, dob.getLength()));
                }
            };
        }
    };
    shuffleHandler.init(conf);
    try {
        shuffleHandler.start();
        DataOutputBuffer outputBuffer = new DataOutputBuffer();
        outputBuffer.reset();
        Token<JobTokenIdentifier> jt = new Token<JobTokenIdentifier>("identifier".getBytes(), "password".getBytes(), new Text(user), new Text("shuffleService"));
        jt.write(outputBuffer);
        shuffleHandler.initializeApplication(new ApplicationInitializationContext(user, appId, ByteBuffer.wrap(outputBuffer.getData(), 0, outputBuffer.getLength())));
        URL url = new URL("http://127.0.0.1:" + shuffleHandler.getConfig().get(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY) + "/mapOutput?job=job_12345_0001&reduce=" + reducerId + "&map=attempt_12345_1_m_1_0");
        HttpURLConnection conn = (HttpURLConnection) url.openConnection();
        conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_NAME, ShuffleHeader.DEFAULT_HTTP_HEADER_NAME);
        conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_VERSION, ShuffleHeader.DEFAULT_HTTP_HEADER_VERSION);
        conn.connect();
        try {
            DataInputStream is = new DataInputStream(conn.getInputStream());
            ShuffleHeader header = new ShuffleHeader();
            header.readFields(is);
            is.close();
        } catch (EOFException e) {
        // ignore
        }
        Assert.assertEquals("sendError called due to shuffle error", 0, failures.size());
    } finally {
        shuffleHandler.stop();
        FileUtil.fullyDelete(absLogDir);
    }
}

Example 10 with DataOutputBuffer

use of org.apache.hadoop.io.DataOutputBuffer in project hadoop by apache.

the class TestShuffleHandler method testClientClosesConnection.

/**
   * Verify client prematurely closing a connection.
   *
   * @throws Exception exception.
   */
@Test(timeout = 10000)
public void testClientClosesConnection() throws Exception {
    final ArrayList<Throwable> failures = new ArrayList<Throwable>(1);
    Configuration conf = new Configuration();
    conf.setInt(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY, 0);
    ShuffleHandler shuffleHandler = new ShuffleHandler() {

        @Override
        protected Shuffle getShuffle(Configuration conf) {
            // replace the shuffle handler with one stubbed for testing
            return new Shuffle(conf) {

                @Override
                protected MapOutputInfo getMapOutputInfo(String mapId, int reduce, String jobId, String user) throws IOException {
                    return null;
                }

                @Override
                protected void populateHeaders(List<String> mapIds, String jobId, String user, int reduce, HttpRequest request, HttpResponse response, boolean keepAliveParam, Map<String, MapOutputInfo> infoMap) throws IOException {
                    // Only set response headers and skip everything else
                    // send some dummy value for content-length
                    super.setResponseHeaders(response, keepAliveParam, 100);
                }

                @Override
                protected void verifyRequest(String appid, ChannelHandlerContext ctx, HttpRequest request, HttpResponse response, URL requestUri) throws IOException {
                }

                @Override
                protected ChannelFuture sendMapOutput(ChannelHandlerContext ctx, Channel ch, String user, String mapId, int reduce, MapOutputInfo info) throws IOException {
                    // send a shuffle header and a lot of data down the channel
                    // to trigger a broken pipe
                    ShuffleHeader header = new ShuffleHeader("attempt_12345_1_m_1_0", 5678, 5678, 1);
                    DataOutputBuffer dob = new DataOutputBuffer();
                    header.write(dob);
                    ch.write(wrappedBuffer(dob.getData(), 0, dob.getLength()));
                    dob = new DataOutputBuffer();
                    for (int i = 0; i < 100000; ++i) {
                        header.write(dob);
                    }
                    return ch.write(wrappedBuffer(dob.getData(), 0, dob.getLength()));
                }

                @Override
                protected void sendError(ChannelHandlerContext ctx, HttpResponseStatus status) {
                    if (failures.size() == 0) {
                        failures.add(new Error());
                        ctx.getChannel().close();
                    }
                }

                @Override
                protected void sendError(ChannelHandlerContext ctx, String message, HttpResponseStatus status) {
                    if (failures.size() == 0) {
                        failures.add(new Error());
                        ctx.getChannel().close();
                    }
                }
            };
        }
    };
    shuffleHandler.init(conf);
    shuffleHandler.start();
    // simulate a reducer that closes early by reading a single shuffle header
    // then closing the connection
    URL url = new URL("http://127.0.0.1:" + shuffleHandler.getConfig().get(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY) + "/mapOutput?job=job_12345_1&reduce=1&map=attempt_12345_1_m_1_0");
    HttpURLConnection conn = (HttpURLConnection) url.openConnection();
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_NAME, ShuffleHeader.DEFAULT_HTTP_HEADER_NAME);
    conn.setRequestProperty(ShuffleHeader.HTTP_HEADER_VERSION, ShuffleHeader.DEFAULT_HTTP_HEADER_VERSION);
    conn.connect();
    DataInputStream input = new DataInputStream(conn.getInputStream());
    Assert.assertEquals(HttpURLConnection.HTTP_OK, conn.getResponseCode());
    Assert.assertEquals("close", conn.getHeaderField(HttpHeader.CONNECTION.asString()));
    ShuffleHeader header = new ShuffleHeader();
    header.readFields(input);
    input.close();
    shuffleHandler.stop();
    Assert.assertTrue("sendError called when client closed connection", failures.size() == 0);
}