protected boolean next(Text key, Text value) throws IOException {
if (fsin.getPos() < end) {
try {
if (readUntilMatch(START_TITLE_MARKER, false)) {
if (readUntilMatch(END_TITLE_MARKER, true)) {
int stop = buffer.getLength() - END_TITLE_MARKER.length;
key.set(buffer.getData(), 0, stop);
buffer.reset();
if (readUntilMatch(START_TEXT_MARKER, false)) {
if (readUntilMatch(END_TEXT_MARKER, true)) {
// un-escape the XML entities encoding and
// re-encode the result as raw UTF8 bytes
stop = buffer.getLength() - END_TITLE_MARKER.length;
String xmlEscapedContent = new String(buffer.getData(), 0, stop + 1, UTF8);
value.set(StringEscapeUtils.unescapeXml(xmlEscapedContent).getBytes(UTF8));
return true;
}
}
}
}
} finally {
buffer.reset();
}
}
return false;
}
private int compareInBinary1(SortComparator comp, ITuple tuple1, ITuple tuple2)
throws IOException {
DataOutputBuffer buffer1 = new DataOutputBuffer();
ser.ser(new DatumWrapper(tuple1), buffer1);
DataOutputBuffer buffer2 = new DataOutputBuffer();
ser.ser(new DatumWrapper(tuple2), buffer2);
return comp.compare(
buffer1.getData(), 0, buffer1.getLength(), buffer2.getData(), 0, buffer2.getLength());
}
/**
* Creates a ByteBuffer with serialized {@link Credentials}.
*
* @param creds The credentials.
* @return The ByteBuffer with the credentials.
* @throws IOException
*/
public static ByteBuffer createTokenBuffer(Credentials creds) throws IOException {
DataOutputBuffer dob = new DataOutputBuffer();
creds.writeTokenStorageToStream(dob);
return ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
}
private void addTimelineDelegationToken(ContainerLaunchContext clc)
throws YarnException, IOException {
Credentials credentials = new Credentials();
DataInputByteBuffer dibb = new DataInputByteBuffer();
ByteBuffer tokens = clc.getTokens();
if (tokens != null) {
dibb.reset(tokens);
credentials.readTokenStorageStream(dibb);
tokens.rewind();
}
// If the timeline delegation token is already in the CLC, no need to add
// one more
for (org.apache.hadoop.security.token.Token<? extends TokenIdentifier> token :
credentials.getAllTokens()) {
if (token.getKind().equals(TimelineDelegationTokenIdentifier.KIND_NAME)) {
return;
}
}
org.apache.hadoop.security.token.Token<TimelineDelegationTokenIdentifier>
timelineDelegationToken = getTimelineDelegationToken();
if (timelineDelegationToken == null) {
return;
}
credentials.addToken(timelineService, timelineDelegationToken);
if (LOG.isDebugEnabled()) {
LOG.debug("Add timline delegation token into credentials: " + timelineDelegationToken);
}
DataOutputBuffer dob = new DataOutputBuffer();
credentials.writeTokenStorageToStream(dob);
tokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
clc.setTokens(tokens);
}
private void requestNewHdfsDelegationToken(
ApplicationId applicationId, String user, boolean shouldCancelAtEnd)
throws IOException, InterruptedException {
// Get new hdfs tokens for this user
Credentials credentials = new Credentials();
Token<?>[] newTokens = obtainSystemTokensForUser(user, credentials);
// Add new tokens to the toRenew list.
LOG.info(
"Received new tokens for " + applicationId + ". Received " + newTokens.length + " tokens.");
if (newTokens.length > 0) {
for (Token<?> token : newTokens) {
if (token.isManaged()) {
DelegationTokenToRenew tokenToRenew =
new DelegationTokenToRenew(
applicationId, token, getConfig(), Time.now(), shouldCancelAtEnd, user);
// renew the token to get the next expiration date.
renewToken(tokenToRenew);
setTimerForTokenRenewal(tokenToRenew);
appTokens.get(applicationId).add(tokenToRenew);
LOG.info("Received new token " + token);
}
}
}
DataOutputBuffer dob = new DataOutputBuffer();
credentials.writeTokenStorageToStream(dob);
ByteBuffer byteBuffer = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
rmContext.getSystemCredentialsForApps().put(applicationId, byteBuffer);
}
/* Initiates a call by sending the parameter to the remote server.
* Note: this is not called from the Connection thread, but by other
* threads.
*/
protected void sendParam(Call call) {
if (shouldCloseConnection.get()) {
return;
}
// For serializing the data to be written.
final DataOutputBuffer d = new DataOutputBuffer();
try {
if (LOG.isDebugEnabled()) LOG.debug(getName() + " sending #" + call.id);
d.writeInt(0xdeadbeef); // placeholder for data length
d.writeInt(call.id);
call.param.write(d);
byte[] data = d.getData();
int dataLength = d.getLength();
// fill in the placeholder
Bytes.putInt(data, 0, dataLength - 4);
//noinspection SynchronizeOnNonFinalField
synchronized (this.out) { // FindBugs IS2_INCONSISTENT_SYNC
out.write(data, 0, dataLength);
out.flush();
}
} catch (IOException e) {
markClosed(e);
} finally {
// the buffer is just an in-memory buffer, but it is still polite to
// close early
IOUtils.closeStream(d);
}
}
@Override
protected void flushAndSync(boolean durable) throws IOException {
int numReadyBytes = buf.countReadyBytes();
if (numReadyBytes > 0) {
int numReadyTxns = buf.countReadyTxns();
long firstTxToFlush = buf.getFirstReadyTxId();
assert numReadyTxns > 0;
// Copy from our double-buffer into a new byte array. This is for
// two reasons:
// 1) The IPC code has no way of specifying to send only a slice of
// a larger array.
// 2) because the calls to the underlying nodes are asynchronous, we
// need a defensive copy to avoid accidentally mutating the buffer
// before it is sent.
DataOutputBuffer bufToSend = new DataOutputBuffer(numReadyBytes);
buf.flushTo(bufToSend);
assert bufToSend.getLength() == numReadyBytes;
byte[] data = bufToSend.getData();
assert data.length == bufToSend.getLength();
QuorumCall<AsyncLogger, Void> qcall =
loggers.sendEdits(
segmentTxId, firstTxToFlush,
numReadyTxns, data);
loggers.waitForWriteQuorum(qcall, writeTimeoutMs, "sendEdits");
// Since we successfully wrote this batch, let the loggers know. Any future
// RPCs will thus let the loggers know of the most recent transaction, even
// if a logger has fallen behind.
loggers.setCommittedTxId(firstTxToFlush + numReadyTxns - 1);
}
}
public static ByteBuffer serializeServiceData(Token<JobTokenIdentifier> jobToken)
throws IOException {
// TODO these bytes should be versioned
DataOutputBuffer jobToken_dob = new DataOutputBuffer();
jobToken.write(jobToken_dob);
return ByteBuffer.wrap(jobToken_dob.getData(), 0, jobToken_dob.getLength());
}
public boolean next(LongWritable key, Text value) throws IOException {
if (pos < end) {
if (readUntilMatch(startTag, false)) {
recordStartPos = pos - startTag.length;
try {
buffer.write(startTag);
if (readUntilMatch(endTag, true)) {
key.set(recordStartPos);
value.set(buffer.getData(), 0, buffer.getLength());
return true;
}
} finally {
// Because input streams of gzipped files are not
// seekable (specifically, do not support getPos), we
// need to keep track of bytes consumed ourselves.
// This is a sanity check to make sure our internal
// computation of bytes consumed is accurate. This
// should be removed later for efficiency once we
// confirm that this code works correctly.
if (fsin instanceof Seekable) {
if (pos != ((Seekable) fsin).getPos()) {
throw new RuntimeException("bytes consumed error!");
}
}
buffer.reset();
}
}
}
return false;
}
private static void rawValueToTextBytes(
DataOutputBuffer dataBuffer, DataInputBuffer inputBuffer, TextBytes textOut)
throws IOException {
inputBuffer.reset(dataBuffer.getData(), dataBuffer.getLength());
int newLength = WritableUtils.readVInt(inputBuffer);
textOut.set(inputBuffer.getData(), inputBuffer.getPosition(), newLength);
}
public static void setTokensFor(
ContainerLaunchContext amContainer, List<Path> paths, Configuration conf) throws IOException {
Credentials credentials = new Credentials();
// for HDFS
TokenCache.obtainTokensForNamenodes(credentials, paths.toArray(new Path[0]), conf);
// for HBase
obtainTokenForHBase(credentials, conf);
// for user
UserGroupInformation currUsr = UserGroupInformation.getCurrentUser();
Collection<Token<? extends TokenIdentifier>> usrTok = currUsr.getTokens();
for (Token<? extends TokenIdentifier> token : usrTok) {
final Text id = new Text(token.getIdentifier());
LOG.info("Adding user token " + id + " with " + token);
credentials.addToken(id, token);
}
try (DataOutputBuffer dob = new DataOutputBuffer()) {
credentials.writeTokenStorageToStream(dob);
if (LOG.isDebugEnabled()) {
LOG.debug("Wrote tokens. Credentials buffer length: " + dob.getLength());
}
ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
amContainer.setTokens(securityTokens);
}
}
/**
* input/output simple records.
*
* @throws Exception if test was failed
*/
@SuppressWarnings("unchecked")
@Test
public void simple_record() throws Exception {
ModelLoader loader = generate();
Class<?> type = loader.modelType("Simple");
assertThat(type.isAnnotationPresent(ModelInputLocation.class), is(true));
assertThat(type.isAnnotationPresent(ModelOutputLocation.class), is(true));
ModelWrapper object = loader.newModel("Simple");
DataOutputBuffer output = new DataOutputBuffer();
ModelOutput<Object> modelOut =
(ModelOutput<Object>)
type.getAnnotation(ModelOutputLocation.class)
.value()
.getDeclaredConstructor(RecordEmitter.class)
.newInstance(new TsvEmitter(new OutputStreamWriter(output, "UTF-8")));
object.set("sid", 1L);
object.set("value", new Text("hello"));
modelOut.write(object.unwrap());
object.set("sid", 2L);
object.set("value", new Text("world"));
modelOut.write(object.unwrap());
object.set("sid", 3L);
object.set("value", null);
modelOut.write(object.unwrap());
modelOut.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), output.getLength());
ModelInput<Object> modelIn =
(ModelInput<Object>)
type.getAnnotation(ModelInputLocation.class)
.value()
.getDeclaredConstructor(RecordParser.class)
.newInstance(new TsvParser(new InputStreamReader(input, "UTF-8")));
ModelWrapper copy = loader.newModel("Simple");
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 1L));
assertThat(copy.get("value"), is((Object) new Text("hello")));
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 2L));
assertThat(copy.get("value"), is((Object) new Text("world")));
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 3L));
assertThat(copy.getOption("value").isNull(), is(true));
assertThat(input.read(), is(-1));
modelIn.close();
}
static byte[] write(Writable writable) {
DataOutputBuffer buffer = new DataOutputBuffer();
buffer.reset();
try {
writable.write(buffer);
} catch (IOException e) {
throw new AssertionError(e);
}
return Arrays.copyOf(buffer.getData(), buffer.getLength());
}
/* Write the header for each connection
* Out is not synchronized because only the first thread does this.
*/
private void writeHeader() throws IOException {
out.write(HBaseServer.HEADER.array());
out.write(HBaseServer.CURRENT_VERSION);
// When there are more fields we can have ConnectionHeader Writable.
DataOutputBuffer buf = new DataOutputBuffer();
header.write(buf);
int bufLen = buf.getLength();
out.writeInt(bufLen);
out.write(buf.getData(), 0, bufLen);
}
@Deprecated
private byte[] getBytes(HRegionInfo hri) throws IOException {
DataOutputBuffer out = new DataOutputBuffer();
try {
hri.write(out);
return out.getData();
} finally {
if (out != null) {
out.close();
}
}
}
/**
* all primitive types.
*
* @throws Exception if test was failed
*/
@SuppressWarnings("unchecked")
@Test
public void primitives() throws Exception {
ModelLoader loader = generate();
Class<?> type = loader.modelType("Primitives");
assertThat(type.isAnnotationPresent(ModelInputLocation.class), is(true));
assertThat(type.isAnnotationPresent(ModelOutputLocation.class), is(true));
ModelWrapper object = loader.newModel("Primitives");
object.set("type_boolean", true);
object.set("type_byte", (byte) 64);
object.set("type_short", (short) 256);
object.set("type_int", 100);
object.set("type_long", 200L);
object.set("type_float", 300.f);
object.set("type_double", 400.d);
object.set("type_decimal", new BigDecimal("1234.567"));
object.set("type_text", new Text("Hello, world!"));
object.set("type_date", new Date(2011, 3, 31));
object.set("type_datetime", new DateTime(2011, 3, 31, 23, 30, 1));
DataOutputBuffer output = new DataOutputBuffer();
ModelOutput<Object> modelOut =
(ModelOutput<Object>)
type.getAnnotation(ModelOutputLocation.class)
.value()
.getDeclaredConstructor(RecordEmitter.class)
.newInstance(new TsvEmitter(new OutputStreamWriter(output, "UTF-8")));
modelOut.write(object.unwrap());
modelOut.write(object.unwrap());
modelOut.write(object.unwrap());
modelOut.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), output.getLength());
ModelInput<Object> modelIn =
(ModelInput<Object>)
type.getAnnotation(ModelInputLocation.class)
.value()
.getDeclaredConstructor(RecordParser.class)
.newInstance(new TsvParser(new InputStreamReader(input, "UTF-8")));
ModelWrapper copy = loader.newModel("Primitives");
modelIn.readTo(copy.unwrap());
assertThat(object.unwrap(), equalTo(copy.unwrap()));
assertThat(input.read(), is(-1));
modelIn.close();
}
protected static void assertSerializable(HadoopSerialization ser, ITuple tuple, boolean debug)
throws IOException {
DataInputBuffer input = new DataInputBuffer();
DataOutputBuffer output = new DataOutputBuffer();
DatumWrapper<ITuple> wrapper = new DatumWrapper<ITuple>(tuple);
ser.ser(wrapper, output);
input.reset(output.getData(), 0, output.getLength());
DatumWrapper<ITuple> wrapper2 = new DatumWrapper<ITuple>();
wrapper2 = ser.deser(wrapper2, input);
if (debug) {
System.out.println("D:" + wrapper2.datum());
}
assertEquals(tuple, wrapper2.datum());
}
/** Used by child copy constructors. */
protected synchronized void copy(Writable other) {
if (other != null) {
try {
DataOutputBuffer out = new DataOutputBuffer();
other.write(out);
DataInputBuffer in = new DataInputBuffer();
in.reset(out.getData(), out.getLength());
readFields(in);
} catch (IOException e) {
throw new IllegalArgumentException("map cannot be copied: " + e.getMessage());
}
} else {
throw new IllegalArgumentException("source map cannot be null");
}
}
@Test
public void testReadWriteReplicaState() {
try {
DataOutputBuffer out = new DataOutputBuffer();
DataInputBuffer in = new DataInputBuffer();
for (HdfsServerConstants.ReplicaState repState : HdfsServerConstants.ReplicaState.values()) {
repState.write(out);
in.reset(out.getData(), out.getLength());
HdfsServerConstants.ReplicaState result = HdfsServerConstants.ReplicaState.read(in);
assertTrue("testReadWrite error !!!", repState == result);
out.reset();
in.reset();
}
} catch (Exception ex) {
fail("testReadWrite ex error ReplicaState");
}
}
@Override
public boolean nextKeyValue() throws IOException, InterruptedException {
if (fsin.getPos() < end) {
if (readUntilMatch(startTag, false)) {
try {
buffer.write(startTag);
if (readUntilMatch(endTag, true)) {
key.set(fsin.getPos());
value.set(buffer.getData(), 0, buffer.getLength());
return true;
}
} finally {
buffer.reset();
}
}
}
return false;
}
public int read() throws IOException {
int ret;
if (null == inbuf || -1 == (ret = inbuf.read())) {
if (!r.next(key, val)) {
return -1;
}
byte[] tmp = key.toString().getBytes();
outbuf.write(tmp, 0, tmp.length);
outbuf.write('\t');
tmp = val.toString().getBytes();
outbuf.write(tmp, 0, tmp.length);
outbuf.write('\n');
inbuf.reset(outbuf.getData(), outbuf.getLength());
outbuf.reset();
ret = inbuf.read();
}
return ret;
}
protected static void assertSerializable(
TupleSerializer ser, TupleDeserializer deser, DatumWrapper<ITuple> tuple, boolean debug)
throws IOException {
DataOutputBuffer output = new DataOutputBuffer();
ser.open(output);
ser.serialize(tuple);
ser.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), 0, output.getLength());
DatumWrapper<ITuple> deserializedTuple = new DatumWrapper<ITuple>();
deser.open(input);
deserializedTuple = deser.deserialize(deserializedTuple);
deser.close();
if (debug) {
System.out.println("D:" + deserializedTuple.datum());
}
assertEquals(tuple.datum(), deserializedTuple.datum());
}
private ByteBuffer getSecurityTokens() throws IOException {
Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials();
Closer closer = Closer.create();
try {
DataOutputBuffer dataOutputBuffer = closer.register(new DataOutputBuffer());
credentials.writeTokenStorageToStream(dataOutputBuffer);
// Remove the AM->RM token so that containers cannot access it
Iterator<Token<?>> tokenIterator = credentials.getAllTokens().iterator();
while (tokenIterator.hasNext()) {
Token<?> token = tokenIterator.next();
if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) {
tokenIterator.remove();
}
}
return ByteBuffer.wrap(dataOutputBuffer.getData(), 0, dataOutputBuffer.getLength());
} catch (Throwable t) {
throw closer.rethrow(t);
} finally {
closer.close();
}
}
@Test(timeout = 5000)
public void testGetBytePayload() throws IOException {
int numBuckets = 10;
VertexManagerPluginContext context = mock(VertexManagerPluginContext.class);
CustomVertexConfiguration vertexConf =
new CustomVertexConfiguration(numBuckets, TezWork.VertexType.INITIALIZED_EDGES);
DataOutputBuffer dob = new DataOutputBuffer();
vertexConf.write(dob);
UserPayload payload = UserPayload.create(ByteBuffer.wrap(dob.getData()));
when(context.getUserPayload()).thenReturn(payload);
CustomPartitionVertex vm = new CustomPartitionVertex(context);
vm.initialize();
// prepare empty routing table
Multimap<Integer, Integer> routingTable = HashMultimap.<Integer, Integer>create();
payload = vm.getBytePayload(routingTable);
// get conf from user payload
CustomEdgeConfiguration edgeConf = new CustomEdgeConfiguration();
DataInputByteBuffer dibb = new DataInputByteBuffer();
dibb.reset(payload.getPayload());
edgeConf.readFields(dibb);
assertEquals(numBuckets, edgeConf.getNumBuckets());
}
/*
* Helper function to create an edge property from an edge type.
*/
private EdgeProperty createEdgeProperty(TezEdgeProperty edgeProp) throws IOException {
DataMovementType dataMovementType;
Class logicalInputClass;
Class logicalOutputClass;
EdgeProperty edgeProperty = null;
EdgeType edgeType = edgeProp.getEdgeType();
switch (edgeType) {
case BROADCAST_EDGE:
dataMovementType = DataMovementType.BROADCAST;
logicalOutputClass = OnFileUnorderedKVOutput.class;
logicalInputClass = ShuffledUnorderedKVInput.class;
break;
case CUSTOM_EDGE:
dataMovementType = DataMovementType.CUSTOM;
logicalOutputClass = OnFileUnorderedPartitionedKVOutput.class;
logicalInputClass = ShuffledUnorderedKVInput.class;
EdgeManagerDescriptor edgeDesc =
new EdgeManagerDescriptor(CustomPartitionEdge.class.getName());
CustomEdgeConfiguration edgeConf =
new CustomEdgeConfiguration(edgeProp.getNumBuckets(), null);
DataOutputBuffer dob = new DataOutputBuffer();
edgeConf.write(dob);
byte[] userPayload = dob.getData();
edgeDesc.setUserPayload(userPayload);
edgeProperty =
new EdgeProperty(
edgeDesc,
DataSourceType.PERSISTED,
SchedulingType.SEQUENTIAL,
new OutputDescriptor(logicalOutputClass.getName()),
new InputDescriptor(logicalInputClass.getName()));
break;
case CUSTOM_SIMPLE_EDGE:
dataMovementType = DataMovementType.SCATTER_GATHER;
logicalOutputClass = OnFileUnorderedPartitionedKVOutput.class;
logicalInputClass = ShuffledUnorderedKVInput.class;
break;
case SIMPLE_EDGE:
default:
dataMovementType = DataMovementType.SCATTER_GATHER;
logicalOutputClass = OnFileSortedOutput.class;
logicalInputClass = ShuffledMergedInputLegacy.class;
break;
}
if (edgeProperty == null) {
edgeProperty =
new EdgeProperty(
dataMovementType,
DataSourceType.PERSISTED,
SchedulingType.SEQUENTIAL,
new OutputDescriptor(logicalOutputClass.getName()),
new InputDescriptor(logicalInputClass.getName()));
}
return edgeProperty;
}
@Override
// Read the page header
// -1: EOF
// 1 - outside the <page> tag
// 2 - just passed the <page> tag but outside the <title>
// 3 - just passed the <title> tag
// 4 - just passed the </title> tag but outside the <namespace>
// 5 - just passed the <namespace>
// 6 - just passed the </namespace> but outside the <id>
// 7 - just passed the (page's) <id>
// 8 - just passed the </id> tag but outside the <revision>
// 9 - (optionally) just passed the <redirect>
// 10 - just passed the (next) <revision>
protected Ack readToPageHeader(RevisionHeader meta) throws IOException {
int i = 0;
int flag = 2;
boolean skipped = false;
int revOrRedirect = -1;
try (DataOutputBuffer pageTitle = new DataOutputBuffer();
DataOutputBuffer nsBuf = new DataOutputBuffer();
DataOutputBuffer keyBuf = new DataOutputBuffer()) {
while (true) {
if (!fetchMore()) return Ack.EOF;
while (hasData()) {
byte b = nextByte();
// when passing the namespace and we realize that
// this is not an article, and that the option of skipping
// non-article pages is on, we simply skip everything until
// the closing </page>
if (skipped) {
if (flag >= 6) {
Log.warn("Peculiar read after skipping namespace");
/*
if (b == END_PAGE[i]) {
i++;
} else i = 0;
if (i >= END_PAGE.length) {
return Ack.SKIPPED;
} */
return Ack.FAILED;
} else return Ack.SKIPPED;
}
if (flag == 2) {
if (b == START_TITLE[i]) {
i++;
} else i = 0;
if (i >= START_TITLE.length) {
flag = 3;
i = 0;
}
}
// put everything between <title></title> block into title
else if (flag == 3) {
if (b == END_TITLE[i]) {
i++;
} else i = 0;
pageTitle.write(b);
if (i >= END_TITLE.length) {
flag = 4;
String title =
new String(pageTitle.getData(), 0, pageTitle.getLength() - END_TITLE.length);
meta.setPageTitle(title);
pageTitle.reset();
i = 0;
}
} else if (flag == 4) {
if (b == START_NAMESPACE[i]) {
i++;
} else i = 0;
if (i >= START_NAMESPACE.length) {
flag = 5;
i = 0;
}
} else if (flag == 5) {
if (b == END_NAMESPACE[i]) {
i++;
} else i = 0;
nsBuf.write(b);
if (i >= END_NAMESPACE.length) {
flag = 6;
String nsStr =
new String(nsBuf.getData(), 0, nsBuf.getLength() - END_NAMESPACE.length);
int ns = Integer.parseInt(nsStr);
nsBuf.reset();
if (ns != 0) {
if (skipNonArticles) {
skipped = true;
meta.clear();
return Ack.SKIPPED;
}
}
meta.setNamespace(ns);
i = 0;
}
} else if (flag == 6) {
if (b == START_ID[i]) {
i++;
} else i = 0;
if (i >= START_ID.length) {
flag = 7;
i = 0;
}
}
// put everything in outer <id></id> block into keyBuf
else if (flag == 7) {
if (b == END_ID[i]) {
i++;
} else i = 0;
keyBuf.write(b);
if (i >= END_ID.length) {
flag = 8;
String idStr = new String(keyBuf.getData(), 0, keyBuf.getLength() - END_ID.length);
long pageId = Long.parseLong(idStr);
meta.setPageId(pageId);
i = 0;
}
} else if (flag == 8) {
int curMatch = 0;
if ((i < START_REVISION.length && b == START_REVISION[i])
&& (i < START_REDIRECT.length && b == START_REDIRECT[i])
// subtle bug here: some tag names can overlap
// multiple times
&& (revOrRedirect == 3 || revOrRedirect == -1)) {
curMatch = 3;
} else if (i < START_REVISION.length && b == START_REVISION[i] && revOrRedirect != 2) {
curMatch = 1;
} else if (i < START_REDIRECT.length && b == START_REDIRECT[i] && revOrRedirect != 1) {
curMatch = 2;
} else {
curMatch = 0;
}
if (curMatch > 0 && (i == 0 || revOrRedirect == 3 || curMatch == revOrRedirect)) {
i++;
revOrRedirect = curMatch;
} else i = 0;
if ((revOrRedirect == 2 || revOrRedirect == 3) && i >= START_REDIRECT.length) {
if (skipRedirect) {
skipped = true;
meta.clear();
return Ack.SKIPPED;
}
revOrRedirect = -1;
flag = 9;
i = 0;
} else if ((revOrRedirect == 1 || revOrRedirect == 3) && i >= START_REVISION.length) {
flag = 10;
revOrRedirect = -1;
return Ack.PASSED_TO_NEXT_TAG;
}
} else if (flag == 9 && !skipRedirect) {
if (b == START_REVISION[i]) {
i++;
} else i = 0;
if (i >= START_REVISION.length) {
flag = 10;
return Ack.PASSED_TO_NEXT_TAG;
}
}
}
}
}
}
/**
* create inmemory segments
*
* @return
* @throws IOException
*/
public List<TezMerger.Segment> createInMemStreams() throws IOException {
int numberOfStreams = Math.max(2, rnd.nextInt(10));
LOG.info("No of streams : " + numberOfStreams);
SerializationFactory serializationFactory = new SerializationFactory(conf);
Serializer keySerializer = serializationFactory.getSerializer(keyClass);
Serializer valueSerializer = serializationFactory.getSerializer(valClass);
LocalDirAllocator localDirAllocator =
new LocalDirAllocator(TezRuntimeFrameworkConfigs.LOCAL_DIRS);
InputContext context = createTezInputContext();
MergeManager mergeManager =
new MergeManager(
conf,
fs,
localDirAllocator,
context,
null,
null,
null,
null,
null,
1024 * 1024 * 10,
null,
false,
-1);
DataOutputBuffer keyBuf = new DataOutputBuffer();
DataOutputBuffer valBuf = new DataOutputBuffer();
DataInputBuffer keyIn = new DataInputBuffer();
DataInputBuffer valIn = new DataInputBuffer();
keySerializer.open(keyBuf);
valueSerializer.open(valBuf);
List<TezMerger.Segment> segments = new LinkedList<TezMerger.Segment>();
for (int i = 0; i < numberOfStreams; i++) {
BoundedByteArrayOutputStream bout = new BoundedByteArrayOutputStream(1024 * 1024);
InMemoryWriter writer = new InMemoryWriter(bout);
Map<Writable, Writable> data = createData();
// write data
for (Map.Entry<Writable, Writable> entry : data.entrySet()) {
keySerializer.serialize(entry.getKey());
valueSerializer.serialize(entry.getValue());
keyIn.reset(keyBuf.getData(), 0, keyBuf.getLength());
valIn.reset(valBuf.getData(), 0, valBuf.getLength());
writer.append(keyIn, valIn);
originalData.put(entry.getKey(), entry.getValue());
keyBuf.reset();
valBuf.reset();
keyIn.reset();
valIn.reset();
}
IFile.Reader reader =
new InMemoryReader(mergeManager, null, bout.getBuffer(), 0, bout.getBuffer().length);
segments.add(new TezMerger.Segment(reader, true));
data.clear();
writer.close();
}
return segments;
}
public void testBinary() throws IOException, InterruptedException {
Configuration conf = new Configuration();
Job job = new Job(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);
}
/**
* Launch application for the dag represented by this client.
*
* @throws YarnException
* @throws IOException
*/
public void startApplication() throws YarnException, IOException {
Class<?>[] defaultClasses;
if (applicationType.equals(YARN_APPLICATION_TYPE)) {
// TODO restrict the security check to only check if security is enabled for webservices.
if (UserGroupInformation.isSecurityEnabled()) {
defaultClasses = DATATORRENT_SECURITY_CLASSES;
} else {
defaultClasses = DATATORRENT_CLASSES;
}
} else {
throw new IllegalStateException(applicationType + " is not a valid application type.");
}
LinkedHashSet<String> localJarFiles = findJars(dag, defaultClasses);
if (resources != null) {
localJarFiles.addAll(resources);
}
YarnClusterMetrics clusterMetrics = yarnClient.getYarnClusterMetrics();
LOG.info(
"Got Cluster metric info from ASM"
+ ", numNodeManagers="
+ clusterMetrics.getNumNodeManagers());
// GetClusterNodesRequest clusterNodesReq = Records.newRecord(GetClusterNodesRequest.class);
// GetClusterNodesResponse clusterNodesResp =
// rmClient.clientRM.getClusterNodes(clusterNodesReq);
// LOG.info("Got Cluster node info from ASM");
// for (NodeReport node : clusterNodesResp.getNodeReports()) {
// LOG.info("Got node report from ASM for"
// + ", nodeId=" + node.getNodeId()
// + ", nodeAddress" + node.getHttpAddress()
// + ", nodeRackName" + node.getRackName()
// + ", nodeNumContainers" + node.getNumContainers()
// + ", nodeHealthStatus" + node.getHealthReport());
// }
List<QueueUserACLInfo> listAclInfo = yarnClient.getQueueAclsInfo();
for (QueueUserACLInfo aclInfo : listAclInfo) {
for (QueueACL userAcl : aclInfo.getUserAcls()) {
LOG.info(
"User ACL Info for Queue"
+ ", queueName="
+ aclInfo.getQueueName()
+ ", userAcl="
+ userAcl.name());
}
}
// Get a new application id
YarnClientApplication newApp = yarnClient.createApplication();
appId = newApp.getNewApplicationResponse().getApplicationId();
// Dump out information about cluster capability as seen by the resource manager
int maxMem = newApp.getNewApplicationResponse().getMaximumResourceCapability().getMemory();
LOG.info("Max mem capabililty of resources in this cluster " + maxMem);
int amMemory = dag.getMasterMemoryMB();
if (amMemory > maxMem) {
LOG.info(
"AM memory specified above max threshold of cluster. Using max value."
+ ", specified="
+ amMemory
+ ", max="
+ maxMem);
amMemory = maxMem;
}
if (dag.getAttributes().get(LogicalPlan.APPLICATION_ID) == null) {
dag.setAttribute(LogicalPlan.APPLICATION_ID, appId.toString());
}
// Create launch context for app master
LOG.info("Setting up application submission context for ASM");
ApplicationSubmissionContext appContext = Records.newRecord(ApplicationSubmissionContext.class);
// set the application id
appContext.setApplicationId(appId);
// set the application name
appContext.setApplicationName(dag.getValue(LogicalPlan.APPLICATION_NAME));
appContext.setApplicationType(this.applicationType);
if (YARN_APPLICATION_TYPE.equals(this.applicationType)) {
// appContext.setMaxAppAttempts(1); // no retries until Stram is HA
}
// Set up the container launch context for the application master
ContainerLaunchContext amContainer = Records.newRecord(ContainerLaunchContext.class);
// Setup security tokens
// If security is enabled get ResourceManager and NameNode delegation tokens.
// Set these tokens on the container so that they are sent as part of application submission.
// This also sets them up for renewal by ResourceManager. The NameNode delegation rmToken
// is also used by ResourceManager to fetch the jars from HDFS and set them up for the
// application master launch.
if (UserGroupInformation.isSecurityEnabled()) {
Credentials credentials = new Credentials();
String tokenRenewer = conf.get(YarnConfiguration.RM_PRINCIPAL);
if (tokenRenewer == null || tokenRenewer.length() == 0) {
throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer");
}
// For now, only getting tokens for the default file-system.
FileSystem fs = StramClientUtils.newFileSystemInstance(conf);
try {
final Token<?> tokens[] = fs.addDelegationTokens(tokenRenewer, credentials);
if (tokens != null) {
for (Token<?> token : tokens) {
LOG.info("Got dt for " + fs.getUri() + "; " + token);
}
}
} finally {
fs.close();
}
addRMDelegationToken(tokenRenewer, credentials);
DataOutputBuffer dob = new DataOutputBuffer();
credentials.writeTokenStorageToStream(dob);
ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
amContainer.setTokens(fsTokens);
}
// set local resources for the application master
// local files or archives as needed
// In this scenario, the jar file for the application master is part of the local resources
Map<String, LocalResource> localResources = new HashMap<String, LocalResource>();
// copy required jar files to dfs, to be localized for containers
FileSystem fs = StramClientUtils.newFileSystemInstance(conf);
try {
Path appsBasePath =
new Path(StramClientUtils.getDTDFSRootDir(fs, conf), StramClientUtils.SUBDIR_APPS);
Path appPath = new Path(appsBasePath, appId.toString());
String libJarsCsv = copyFromLocal(fs, appPath, localJarFiles.toArray(new String[] {}));
LOG.info("libjars: {}", libJarsCsv);
dag.getAttributes().put(LogicalPlan.LIBRARY_JARS, libJarsCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.FILE, libJarsCsv, localResources, fs);
if (archives != null) {
String[] localFiles = archives.split(",");
String archivesCsv = copyFromLocal(fs, appPath, localFiles);
LOG.info("archives: {}", archivesCsv);
dag.getAttributes().put(LogicalPlan.ARCHIVES, archivesCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.ARCHIVE, archivesCsv, localResources, fs);
}
if (files != null) {
String[] localFiles = files.split(",");
String filesCsv = copyFromLocal(fs, appPath, localFiles);
LOG.info("files: {}", filesCsv);
dag.getAttributes().put(LogicalPlan.FILES, filesCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.FILE, filesCsv, localResources, fs);
}
dag.getAttributes().put(LogicalPlan.APPLICATION_PATH, appPath.toString());
if (dag.getAttributes().get(OperatorContext.STORAGE_AGENT) == null) {
/* which would be the most likely case */
Path checkpointPath = new Path(appPath, LogicalPlan.SUBDIR_CHECKPOINTS);
// use conf client side to pickup any proxy settings from dt-site.xml
dag.setAttribute(
OperatorContext.STORAGE_AGENT, new FSStorageAgent(checkpointPath.toString(), conf));
}
if (dag.getAttributes().get(LogicalPlan.CONTAINER_OPTS_CONFIGURATOR) == null) {
dag.setAttribute(
LogicalPlan.CONTAINER_OPTS_CONFIGURATOR, new BasicContainerOptConfigurator());
}
// Set the log4j properties if needed
if (!log4jPropFile.isEmpty()) {
Path log4jSrc = new Path(log4jPropFile);
Path log4jDst = new Path(appPath, "log4j.props");
fs.copyFromLocalFile(false, true, log4jSrc, log4jDst);
FileStatus log4jFileStatus = fs.getFileStatus(log4jDst);
LocalResource log4jRsrc = Records.newRecord(LocalResource.class);
log4jRsrc.setType(LocalResourceType.FILE);
log4jRsrc.setVisibility(LocalResourceVisibility.APPLICATION);
log4jRsrc.setResource(ConverterUtils.getYarnUrlFromURI(log4jDst.toUri()));
log4jRsrc.setTimestamp(log4jFileStatus.getModificationTime());
log4jRsrc.setSize(log4jFileStatus.getLen());
localResources.put("log4j.properties", log4jRsrc);
}
if (originalAppId != null) {
Path origAppPath = new Path(appsBasePath, this.originalAppId);
LOG.info("Restart from {}", origAppPath);
copyInitialState(origAppPath);
}
// push logical plan to DFS location
Path cfgDst = new Path(appPath, LogicalPlan.SER_FILE_NAME);
FSDataOutputStream outStream = fs.create(cfgDst, true);
LogicalPlan.write(this.dag, outStream);
outStream.close();
Path launchConfigDst = new Path(appPath, LogicalPlan.LAUNCH_CONFIG_FILE_NAME);
outStream = fs.create(launchConfigDst, true);
conf.writeXml(outStream);
outStream.close();
FileStatus topologyFileStatus = fs.getFileStatus(cfgDst);
LocalResource topologyRsrc = Records.newRecord(LocalResource.class);
topologyRsrc.setType(LocalResourceType.FILE);
topologyRsrc.setVisibility(LocalResourceVisibility.APPLICATION);
topologyRsrc.setResource(ConverterUtils.getYarnUrlFromURI(cfgDst.toUri()));
topologyRsrc.setTimestamp(topologyFileStatus.getModificationTime());
topologyRsrc.setSize(topologyFileStatus.getLen());
localResources.put(LogicalPlan.SER_FILE_NAME, topologyRsrc);
// Set local resource info into app master container launch context
amContainer.setLocalResources(localResources);
// Set the necessary security tokens as needed
// amContainer.setContainerTokens(containerToken);
// Set the env variables to be setup in the env where the application master will be run
LOG.info("Set the environment for the application master");
Map<String, String> env = new HashMap<String, String>();
// Add application jar(s) location to classpath
// At some point we should not be required to add
// the hadoop specific classpaths to the env.
// It should be provided out of the box.
// For now setting all required classpaths including
// the classpath to "." for the application jar(s)
// including ${CLASSPATH} will duplicate the class path in app master, removing it for now
// StringBuilder classPathEnv = new StringBuilder("${CLASSPATH}:./*");
StringBuilder classPathEnv = new StringBuilder("./*");
String classpath = conf.get(YarnConfiguration.YARN_APPLICATION_CLASSPATH);
for (String c :
StringUtils.isBlank(classpath)
? YarnConfiguration.DEFAULT_YARN_APPLICATION_CLASSPATH
: classpath.split(",")) {
if (c.equals("$HADOOP_CLIENT_CONF_DIR")) {
// SPOI-2501
continue;
}
classPathEnv.append(':');
classPathEnv.append(c.trim());
}
env.put("CLASSPATH", classPathEnv.toString());
// propagate to replace node managers user name (effective in non-secure mode)
env.put("HADOOP_USER_NAME", UserGroupInformation.getLoginUser().getUserName());
amContainer.setEnvironment(env);
// Set the necessary command to execute the application master
ArrayList<CharSequence> vargs = new ArrayList<CharSequence>(30);
// Set java executable command
LOG.info("Setting up app master command");
vargs.add(javaCmd);
if (dag.isDebug()) {
vargs.add("-agentlib:jdwp=transport=dt_socket,server=y,suspend=n");
}
// Set Xmx based on am memory size
// default heap size 75% of total memory
vargs.add("-Xmx" + (amMemory * 3 / 4) + "m");
vargs.add("-XX:+HeapDumpOnOutOfMemoryError");
vargs.add("-XX:HeapDumpPath=/tmp/dt-heap-" + appId.getId() + ".bin");
vargs.add("-Dhadoop.root.logger=" + (dag.isDebug() ? "DEBUG" : "INFO") + ",RFA");
vargs.add("-Dhadoop.log.dir=" + ApplicationConstants.LOG_DIR_EXPANSION_VAR);
vargs.add(String.format("-D%s=%s", StreamingContainer.PROP_APP_PATH, dag.assertAppPath()));
if (dag.isDebug()) {
vargs.add("-Dlog4j.debug=true");
}
String loggersLevel = conf.get(DTLoggerFactory.DT_LOGGERS_LEVEL);
if (loggersLevel != null) {
vargs.add(String.format("-D%s=%s", DTLoggerFactory.DT_LOGGERS_LEVEL, loggersLevel));
}
vargs.add(StreamingAppMaster.class.getName());
vargs.add("1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stdout");
vargs.add("2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stderr");
// Get final command
StringBuilder command = new StringBuilder(9 * vargs.size());
for (CharSequence str : vargs) {
command.append(str).append(" ");
}
LOG.info("Completed setting up app master command " + command.toString());
List<String> commands = new ArrayList<String>();
commands.add(command.toString());
amContainer.setCommands(commands);
// Set up resource type requirements
// For now, only memory is supported so we set memory requirements
Resource capability = Records.newRecord(Resource.class);
capability.setMemory(amMemory);
appContext.setResource(capability);
// Service data is a binary blob that can be passed to the application
// Not needed in this scenario
// amContainer.setServiceData(serviceData);
appContext.setAMContainerSpec(amContainer);
// Set the priority for the application master
Priority pri = Records.newRecord(Priority.class);
pri.setPriority(amPriority);
appContext.setPriority(pri);
// Set the queue to which this application is to be submitted in the RM
appContext.setQueue(queueName);
// Submit the application to the applications manager
// SubmitApplicationResponse submitResp = rmClient.submitApplication(appRequest);
// Ignore the response as either a valid response object is returned on success
// or an exception thrown to denote some form of a failure
String specStr =
Objects.toStringHelper("Submitting application: ")
.add("name", appContext.getApplicationName())
.add("queue", appContext.getQueue())
.add("user", UserGroupInformation.getLoginUser())
.add("resource", appContext.getResource())
.toString();
LOG.info(specStr);
if (dag.isDebug()) {
// LOG.info("Full submission context: " + appContext);
}
yarnClient.submitApplication(appContext);
} finally {
fs.close();
}
}
public void writeRequest(OutputStream out) throws IOException {
DataOutputStream dos = new DataOutputStream(out);
dos.writeInt(buffers.size());
for (DataOutputBuffer b : buffers) dos.write(b.getData(), 0, b.getLength());
}