Search in sources :

Example 26 with DataInputStream

use of java.io.DataInputStream in project hadoop by apache.

the class TestJoinTupleWritable method testSparseWideWritable.

/**
   * Tests a tuple writable with more than 64 values and the values set written
   * spread far apart.
   */
@Test
public void testSparseWideWritable() throws Exception {
    Writable[] manyWrits = makeRandomWritables(131);
    TupleWritable sTuple = new TupleWritable(manyWrits);
    for (int i = 0; i < manyWrits.length; i++) {
        if (i % 65 == 0) {
            sTuple.setWritten(i);
        }
    }
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    sTuple.write(new DataOutputStream(out));
    ByteArrayInputStream in = new ByteArrayInputStream(out.toByteArray());
    TupleWritable dTuple = new TupleWritable();
    dTuple.readFields(new DataInputStream(in));
    assertTrue("Failed to write/read tuple", sTuple.equals(dTuple));
    assertEquals("All tuple data has not been read from the stream", -1, in.read());
}
Also used : ByteArrayInputStream(java.io.ByteArrayInputStream) DataOutputStream(java.io.DataOutputStream) Writable(org.apache.hadoop.io.Writable) LongWritable(org.apache.hadoop.io.LongWritable) BooleanWritable(org.apache.hadoop.io.BooleanWritable) BytesWritable(org.apache.hadoop.io.BytesWritable) FloatWritable(org.apache.hadoop.io.FloatWritable) IntWritable(org.apache.hadoop.io.IntWritable) ByteArrayOutputStream(java.io.ByteArrayOutputStream) DataInputStream(java.io.DataInputStream) Test(org.junit.Test)

Example 27 with DataInputStream

use of java.io.DataInputStream in project hadoop by apache.

the class TestJoinTupleWritable method testWideWritable2.

@Test
public void testWideWritable2() throws Exception {
    Writable[] manyWrits = makeRandomWritables(71);
    TupleWritable sTuple = new TupleWritable(manyWrits);
    for (int i = 0; i < manyWrits.length; i++) {
        sTuple.setWritten(i);
    }
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    sTuple.write(new DataOutputStream(out));
    ByteArrayInputStream in = new ByteArrayInputStream(out.toByteArray());
    TupleWritable dTuple = new TupleWritable();
    dTuple.readFields(new DataInputStream(in));
    assertTrue("Failed to write/read tuple", sTuple.equals(dTuple));
    assertEquals("All tuple data has not been read from the stream", -1, in.read());
}
Also used : ByteArrayInputStream(java.io.ByteArrayInputStream) DataOutputStream(java.io.DataOutputStream) Writable(org.apache.hadoop.io.Writable) LongWritable(org.apache.hadoop.io.LongWritable) BooleanWritable(org.apache.hadoop.io.BooleanWritable) BytesWritable(org.apache.hadoop.io.BytesWritable) FloatWritable(org.apache.hadoop.io.FloatWritable) IntWritable(org.apache.hadoop.io.IntWritable) ByteArrayOutputStream(java.io.ByteArrayOutputStream) DataInputStream(java.io.DataInputStream) Test(org.junit.Test)

Example 28 with DataInputStream

use of java.io.DataInputStream in project hadoop by apache.

the class TestJoinTupleWritable method testWritable.

@Test
public void testWritable() throws Exception {
    Random r = new Random();
    Writable[] writs = { new BooleanWritable(r.nextBoolean()), new FloatWritable(r.nextFloat()), new FloatWritable(r.nextFloat()), new IntWritable(r.nextInt()), new LongWritable(r.nextLong()), new BytesWritable("dingo".getBytes()), new LongWritable(r.nextLong()), new IntWritable(r.nextInt()), new BytesWritable("yak".getBytes()), new IntWritable(r.nextInt()) };
    TupleWritable sTuple = makeTuple(writs);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    sTuple.write(new DataOutputStream(out));
    ByteArrayInputStream in = new ByteArrayInputStream(out.toByteArray());
    TupleWritable dTuple = new TupleWritable();
    dTuple.readFields(new DataInputStream(in));
    assertTrue("Failed to write/read tuple", sTuple.equals(dTuple));
}
Also used : DataOutputStream(java.io.DataOutputStream) Writable(org.apache.hadoop.io.Writable) LongWritable(org.apache.hadoop.io.LongWritable) BooleanWritable(org.apache.hadoop.io.BooleanWritable) BytesWritable(org.apache.hadoop.io.BytesWritable) FloatWritable(org.apache.hadoop.io.FloatWritable) IntWritable(org.apache.hadoop.io.IntWritable) BytesWritable(org.apache.hadoop.io.BytesWritable) ByteArrayOutputStream(java.io.ByteArrayOutputStream) DataInputStream(java.io.DataInputStream) FloatWritable(org.apache.hadoop.io.FloatWritable) Random(java.util.Random) BooleanWritable(org.apache.hadoop.io.BooleanWritable) ByteArrayInputStream(java.io.ByteArrayInputStream) LongWritable(org.apache.hadoop.io.LongWritable) IntWritable(org.apache.hadoop.io.IntWritable) Test(org.junit.Test)

Example 29 with DataInputStream

use of java.io.DataInputStream 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();
}
Also used : HttpRequest(org.jboss.netty.handler.codec.http.HttpRequest) Configuration(org.apache.hadoop.conf.Configuration) YarnConfiguration(org.apache.hadoop.yarn.conf.YarnConfiguration) HttpResponseStatus(org.jboss.netty.handler.codec.http.HttpResponseStatus) ShuffleHeader(org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader) SocketChannel(org.jboss.netty.channel.socket.SocketChannel) Channel(org.jboss.netty.channel.Channel) AbstractChannel(org.jboss.netty.channel.AbstractChannel) ArrayList(java.util.ArrayList) DefaultHttpResponse(org.jboss.netty.handler.codec.http.DefaultHttpResponse) HttpResponse(org.jboss.netty.handler.codec.http.HttpResponse) ChannelHandlerContext(org.jboss.netty.channel.ChannelHandlerContext) DataInputStream(java.io.DataInputStream) URL(java.net.URL) HttpURLConnection(java.net.HttpURLConnection) DefaultHttpResponse(org.jboss.netty.handler.codec.http.DefaultHttpResponse) DataOutputBuffer(org.apache.hadoop.io.DataOutputBuffer) List(java.util.List) ArrayList(java.util.ArrayList) Map(java.util.Map) Test(org.junit.Test)

Example 30 with DataInputStream

use of java.io.DataInputStream 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);
    }
}
Also used : Configuration(org.apache.hadoop.conf.Configuration) YarnConfiguration(org.apache.hadoop.yarn.conf.YarnConfiguration) HttpResponseStatus(org.jboss.netty.handler.codec.http.HttpResponseStatus) ArrayList(java.util.ArrayList) ChannelHandlerContext(org.jboss.netty.channel.ChannelHandlerContext) Token(org.apache.hadoop.security.token.Token) URL(java.net.URL) HttpURLConnection(java.net.HttpURLConnection) DataOutputBuffer(org.apache.hadoop.io.DataOutputBuffer) EOFException(java.io.EOFException) List(java.util.List) ArrayList(java.util.ArrayList) ApplicationInitializationContext(org.apache.hadoop.yarn.server.api.ApplicationInitializationContext) HttpRequest(org.jboss.netty.handler.codec.http.HttpRequest) SocketChannel(org.jboss.netty.channel.socket.SocketChannel) Channel(org.jboss.netty.channel.Channel) AbstractChannel(org.jboss.netty.channel.AbstractChannel) ShuffleHeader(org.apache.hadoop.mapreduce.task.reduce.ShuffleHeader) DefaultHttpResponse(org.jboss.netty.handler.codec.http.DefaultHttpResponse) HttpResponse(org.jboss.netty.handler.codec.http.HttpResponse) JobTokenIdentifier(org.apache.hadoop.mapreduce.security.token.JobTokenIdentifier) Text(org.apache.hadoop.io.Text) DataInputStream(java.io.DataInputStream) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) File(java.io.File) Map(java.util.Map) Test(org.junit.Test)

Aggregations

DataInputStream (java.io.DataInputStream)2761 ByteArrayInputStream (java.io.ByteArrayInputStream)1139 IOException (java.io.IOException)1043 DataOutputStream (java.io.DataOutputStream)606 FileInputStream (java.io.FileInputStream)542 Test (org.junit.Test)533 ByteArrayOutputStream (java.io.ByteArrayOutputStream)368 File (java.io.File)274 BufferedInputStream (java.io.BufferedInputStream)253 InputStream (java.io.InputStream)245 ArrayList (java.util.ArrayList)200 EOFException (java.io.EOFException)154 DataInput (java.io.DataInput)141 FileNotFoundException (java.io.FileNotFoundException)131 ByteBuffer (java.nio.ByteBuffer)119 FileOutputStream (java.io.FileOutputStream)105 HashMap (java.util.HashMap)101 BufferedReader (java.io.BufferedReader)90 InputStreamReader (java.io.InputStreamReader)89 Socket (java.net.Socket)75