Examples with ShuffleHeader org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader used on opensource projects

Example 1 with ShuffleHeader

use of org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader 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 2 with ShuffleHeader

use of org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader 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 3 with ShuffleHeader

use of org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader 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);
}

Example 4 with ShuffleHeader

use of org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader in project hadoop by apache.

the class TestShuffleHandler method testMaxConnections.

/**
   * Validate the limit on number of shuffle connections.
   * 
   * @throws Exception exception
   */
@Test(timeout = 10000)
public void testMaxConnections() throws Exception {
    Configuration conf = new Configuration();
    conf.setInt(ShuffleHandler.SHUFFLE_PORT_CONFIG_KEY, 0);
    conf.setInt(ShuffleHandler.MAX_SHUFFLE_CONNECTIONS, 3);
    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 {
                    // Do nothing.
                    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 {
                // Do nothing.
                }

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

                @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("dummy_header", 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()));
                }
            };
        }
    };
    shuffleHandler.init(conf);
    shuffleHandler.start();
    // setup connections
    int connAttempts = 3;
    HttpURLConnection[] conns = new HttpURLConnection[connAttempts];
    for (int i = 0; i < connAttempts; i++) {
        String URLstring = "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_" + i + "_0";
        URL url = new URL(URLstring);
        conns[i] = (HttpURLConnection) url.openConnection();
        conns[i].setRequestProperty(ShuffleHeader.HTTP_HEADER_NAME, ShuffleHeader.DEFAULT_HTTP_HEADER_NAME);
        conns[i].setRequestProperty(ShuffleHeader.HTTP_HEADER_VERSION, ShuffleHeader.DEFAULT_HTTP_HEADER_VERSION);
    }
    // Try to open numerous connections
    for (int i = 0; i < connAttempts; i++) {
        conns[i].connect();
    }
    //Ensure first connections are okay
    conns[0].getInputStream();
    int rc = conns[0].getResponseCode();
    Assert.assertEquals(HttpURLConnection.HTTP_OK, rc);
    conns[1].getInputStream();
    rc = conns[1].getResponseCode();
    Assert.assertEquals(HttpURLConnection.HTTP_OK, rc);
    // This connection should be closed because it to above the limit
    try {
        rc = conns[2].getResponseCode();
        Assert.assertEquals("Expected a too-many-requests response code", ShuffleHandler.TOO_MANY_REQ_STATUS.getCode(), rc);
        long backoff = Long.valueOf(conns[2].getHeaderField(ShuffleHandler.RETRY_AFTER_HEADER));
        Assert.assertTrue("The backoff value cannot be negative.", backoff > 0);
        conns[2].getInputStream();
        Assert.fail("Expected an IOException");
    } catch (IOException ioe) {
        LOG.info("Expected - connection should not be open");
    } catch (NumberFormatException ne) {
        Assert.fail("Expected a numerical value for RETRY_AFTER header field");
    } catch (Exception e) {
        Assert.fail("Expected a IOException");
    }
    shuffleHandler.stop();
}