HBase RPC的client主要工作：
1.JDK动态代理获取代理类
2.调用对应服务方法，逻辑包装在Invoker里，新建连接，发送数据包，等待server端响应，默认超时60s
3.超时使用wait+loop检查实现
其类图如下
0.94实现如下
HBaseRPC getProxy入口
public static VersionedProtocol getProxy(
Class<? extends VersionedProtocol> protocol,
long clientVersion, InetSocketAddress addr, User ticket,
Configuration conf, SocketFactory factory, int rpcTimeout)
throws IOException {
//获取代理类
VersionedProtocol proxy =
getProtocolEngine(protocol,conf)
.getProxy(protocol, clientVersion, addr, ticket, conf, factory, Math.min(rpcTimeout, HBaseRPC.getRpcTimeout()));
//发起rpc调用
long serverVersion = proxy.getProtocolVersion(protocol.getName(),
clientVersion);
if (serverVersion == clientVersion) {
return proxy;
}
throw new VersionMismatch(protocol.getName(), clientVersion,
serverVersion);
}
public VersionedProtocol getProxy(
Class<? extends VersionedProtocol> protocol, long clientVersion,
InetSocketAddress addr, User ticket,
Configuration conf, SocketFactory factory, int rpcTimeout)
throws IOException {
//jdk动态代理
VersionedProtocol proxy =
(VersionedProtocol) Proxy.newProxyInstance(
protocol.getClassLoader(), new Class[] { protocol },
new Invoker(protocol, addr, ticket, conf, factory, rpcTimeout));
if (proxy instanceof VersionedProtocol) {
long serverVersion = ((VersionedProtocol)proxy)
.getProtocolVersion(protocol.getName(), clientVersion);
if (serverVersion != clientVersion) {
throw new HBaseRPC.VersionMismatch(protocol.getName(), clientVersion,
serverVersion);
}
}
return proxy;
}
调用发起 invoker
public Object invoke(Object proxy, Method method, Object[] args)
throws Throwable {
final boolean logDebug = LOG.isDebugEnabled();
long startTime = 0;
if (logDebug) {
startTime = System.currentTimeMillis();
}
//调用
HbaseObjectWritable value = (HbaseObjectWritable)
client.call(new Invocation(method, args), address,
protocol, ticket, rpcTimeout);
if (logDebug) {
// FIGURE HOW TO TURN THIS OFF!
long callTime = System.currentTimeMillis() - startTime;
LOG.debug("Call: " + method.getName() + " " + callTime);
}
return value.get();
}
Invocation组装
public Invocation(Method method, Object[] parameters) {
this.methodName = method.getName();
this.parameterClasses = method.getParameterTypes();
this.parameters = parameters;
if (method.getDeclaringClass().equals(VersionedProtocol.class)) {
//VersionedProtocol is exempted from version check.
clientVersion = 0;
clientMethodsHash = 0;
} else {
try {
Field versionField = method.getDeclaringClass().getField("VERSION");
versionField.setAccessible(true);
this.clientVersion = versionField.getLong(method.getDeclaringClass());
} catch (NoSuchFieldException ex) {
throw new RuntimeException("The " + method.getDeclaringClass(), ex);
} catch (IllegalAccessException ex) {
throw new RuntimeException(ex);
}
this.clientMethodsHash = ProtocolSignature.getFingerprint(method
.getDeclaringClass().getMethods());
}
}
调用过程
public Writable call(Writable param, InetSocketAddress addr,
Class<? extends VersionedProtocol> protocol,
User ticket, int rpcTimeout)
throws InterruptedException, IOException {、
//Call代表一次请求，具有唯一id
Call call = new Call(param);
//获取连接，此处会创建连接，初始化IO操作
Connection connection = getConnection(addr, protocol, ticket, rpcTimeout, call);
//发送请求
connection.sendParam(call); // send the parameter
boolean interrupted = false;
//noinspection SynchronizationOnLocalVariableOrMethodParameter
//业务线程在此等待server端响应
synchronized (call) {
while (!call.done) {
try {
call.wait(); // wait for the result
} catch (InterruptedException ignored) {
// save the fact that we were interrupted
interrupted = true;
}
}
if (interrupted) {
// set the interrupt flag now that we are done waiting
Thread.currentThread().interrupt();
}
//如果有异常，则抛出
if (call.error != null) {
if (call.error instanceof RemoteException) {
call.error.fillInStackTrace();
throw call.error;
}
// local exception
throw wrapException(addr, call.error);
}
//否则，返回调用值
return call.value;
}
}
Call初始化
protected Call(Writable param) {
this.param = param;
this.startTime = System.currentTimeMillis();
//唯一id
synchronized (HBaseClient.this) {
this.id = counter++;
}
}
getConnection操作
protected Connection getConnection(InetSocketAddress addr,
Class<? extends VersionedProtocol> protocol,
User ticket,
int rpcTimeout,
Call call)
throws IOException, InterruptedException {
if (!running.get()) {
// the client is stopped
throw new IOException("The client is stopped");
}
Connection connection;
/* we could avoid this allocation for each RPC by having a
* connectionsId object and with set() method. We need to manage the
* refs for keys in HashMap properly. For now its ok.
*/
//每个连接唯一标示，不同标示会建立不同连接
ConnectionId remoteId = new ConnectionId(addr, protocol, ticket, rpcTimeout);
do {
synchronized (connections) {
connection = connections.get(remoteId);
if (connection == null) {
connection = new Connection(remoteId);
connections.put(remoteId, connection);
}
}
}
//将请求添加到内部Map，方便后续server返回数据时处理
while (!connection.addCall(call));
//we don't invoke the method below inside "synchronized (connections)"
//block above. The reason for that is if the server happens to be slow,
//it will take longer to establish a connection and that will slow the
//entire system down.
//新建连接，初始化IO
connection.setupIOstreams();
return connection;
}
新建连接过程
protected synchronized void setupIOstreams()
throws IOException, InterruptedException {
if (socket != null || shouldCloseConnection.get()) {
return;
}
try {
if (LOG.isDebugEnabled()) {
LOG.debug("Connecting to "+remoteId);
}
//创建连接
setupConnection();
//初始化IO
this.in = new DataInputStream(new BufferedInputStream
(new PingInputStream(NetUtils.getInputStream(socket))));
this.out = new DataOutputStream
(new BufferedOutputStream(NetUtils.getOutputStream(socket)));
//发送header，还在buffer中
writeHeader();
// update last activity time
//最新活动时间
touch();
//启动读线程，该线程监听OS的READ事件，负责从server端读取数据
// start the receiver thread after the socket connection has been set up
start();
} catch (IOException e) {
markClosed(e);
close();
throw e;
}
}
创建连接
protected synchronized void setupConnection() throws IOException {
short ioFailures = 0;
short timeoutFailures = 0;
while (true) {
try {
//默认是socketFactory是StandardSocketFactory
this.socket = socketFactory.createSocket();
this.socket.setTcpNoDelay(tcpNoDelay);
this.socket.setKeepAlive(tcpKeepAlive);
// connection time out is 20s
//连接，直到连上，默认超时20s
NetUtils.connect(this.socket, remoteId.getAddress(),
getSocketTimeout(conf));
if (remoteId.rpcTimeout > 0) {
pingInterval = remoteId.rpcTimeout; // overwrite pingInterval
}
this.socket.setSoTimeout(pingInterval);
return;
} catch (SocketTimeoutException toe) {
/* The max number of retries is 45,
* which amounts to 20s*45 = 15 minutes retries.
*/
handleConnectionFailure(timeoutFailures++, maxRetries, toe);
} catch (IOException ie) {
handleConnectionFailure(ioFailures++, maxRetries, ie);
}
}
}
StandardSocketFactory创建socket
public Socket createSocket() throws IOException {
/*
* NOTE: This returns an NIO socket so that it has an associated
* SocketChannel. As of now, this unfortunately makes streams returned
* by Socket.getInputStream() and Socket.getOutputStream() unusable
* (because a blocking read on input stream blocks write on output stream
* and vice versa).
*
* So users of these socket factories should use
* NetUtils.getInputStream(socket) and
* NetUtils.getOutputStream(socket) instead.
*
* A solution for hiding from this from user is to write a
* 'FilterSocket' on the lines of FilterInputStream and extend it by
* overriding getInputStream() and getOutputStream().
*/
//打开一个channel
return SocketChannel.open().socket();
}
连接过程
public static void connect(Socket socket,
SocketAddress endpoint,
int timeout) throws IOException {
if (socket == null || endpoint == null || timeout < 0) {
throw new IllegalArgumentException("Illegal argument for connect()");
}
SocketChannel ch = socket.getChannel();
if (ch == null) {
// let the default implementation handle it.
socket.connect(endpoint, timeout);
} else {
SocketIOWithTimeout.connect(ch, endpoint, timeout);
}
......
}
static void connect(SocketChannel channel,
SocketAddress endpoint, int timeout) throws IOException {
//使用NIO
boolean blockingOn = channel.isBlocking();
if (blockingOn) {
channel.configureBlocking(false);
}
//尝试连接
try {
if (channel.connect(endpoint)) {
return;
}
//如果还没连上，则开启一个selector，监听CONNECT事件
long timeoutLeft = timeout;
long endTime = (timeout > 0) ? (System.currentTimeMillis() + timeout): 0;
while (true) {
// we might have to call finishConnect() more than once
// for some channels (with user level protocols)
//CONNECT是否就位
int ret = selector.select((SelectableChannel)channel,
SelectionKey.OP_CONNECT, timeoutLeft);
//如果就位，再看看channel是否已连接
if (ret > 0 && channel.finishConnect()) {
return;
}
//如果没就位，或者超时了，抛出异常
if (ret == 0 ||
(timeout > 0 &&
(timeoutLeft = (endTime - System.currentTimeMillis())) <= 0)) {
throw new SocketTimeoutException(
timeoutExceptionString(channel, timeout,
SelectionKey.OP_CONNECT));
}
}
} catch (IOException e) {
// javadoc for SocketChannel.connect() says channel should be closed.
try {
channel.close();
} catch (IOException ignored) {}
throw e;
} finally {
if (blockingOn && channel.isOpen()) {
channel.configureBlocking(true);
}
}
}
再看SelectorPool的select过程
int select(SelectableChannel channel, int ops, long timeout)
throws IOException {
//从pool中拿一个selector，SelectorPool维护着一个Provider列表，每个provider都有一个selector队列
SelectorInfo info = get(channel);
SelectionKey key = null;
int ret = 0;
try {
while (true) {
long start = (timeout == 0) ? 0 : System.currentTimeMillis();
//CONNECT就位事件
key = channel.register(info.selector, ops);
//阻塞select，直到超时
ret = info.selector.select(timeout);
if (ret != 0) {
return ret;
}
/* Sometimes select() returns 0 much before timeout for
* unknown reasons. So select again if required.
*/
//如果超时了，返回0，没超时则继续select
if (timeout > 0) {
timeout -= System.currentTimeMillis() - start;
if (timeout <= 0) {
return 0;
}
}
if (Thread.currentThread().isInterrupted()) {
throw new InterruptedIOException("Interruped while waiting for " +
"IO on channel " + channel +
". " + timeout +
" millis timeout left.");
}
}
} finally {
//处理完后，清理key
if (key != null) {
key.cancel();
}
//clear the canceled key.
try {
info.selector.selectNow();
} catch (IOException e) {
LOG.info("Unexpected Exception while clearing selector : " +
StringUtils.stringifyException(e));
// don't put the selector back.
info.close();
return ret;
}
release(info);
}
}
连上之后，初始化IO
public SocketInputStream(ReadableByteChannel channel, long timeout)
throws IOException {
SocketIOWithTimeout.checkChannelValidity(channel);
reader = new Reader(channel, timeout);
}
SocketIOWithTimeout(SelectableChannel channel, long timeout)
throws IOException {
checkChannelValidity(channel);
this.channel = channel;
this.timeout = timeout;
// Set non-blocking
channel.configureBlocking(false);
}
public SocketOutputStream(WritableByteChannel channel, long timeout)
throws IOException {
SocketIOWithTimeout.checkChannelValidity(channel);
writer = new Writer(channel, timeout);
}
写header
private void writeHeader() throws IOException {
//固定4个字节
out.write(HBaseServer.HEADER.array());
//版本信息，一个字节
out.write(HBaseServer.CURRENT_VERSION);
//When there are more fields we can have ConnectionHeader Writable.
DataOutputBuffer buf = new DataOutputBuffer();
//header序列化byte
header.write(buf);
//写长度，4个字节，写内容
int bufLen = buf.getLength();
out.writeInt(bufLen);
out.write(buf.getData(), 0, bufLen);
}
之后IO READ线程Connection启动
public void run() {
if (LOG.isDebugEnabled())
LOG.debug(getName() + ": starting, having connections "
+ connections.size());
try {
//如果没有请求，则等一段时间
while (waitForWork()) {//wait here for work - read or close connection
receiveResponse();
}
} catch (Throwable t) {
LOG.warn("Unexpected exception receiving call responses", t);
markClosed(new IOException("Unexpected exception receiving call responses", t));
}
close();
if (LOG.isDebugEnabled())
LOG.debug(getName() + ": stopped, remaining connections "
+ connections.size());
}
数据读取
protected void receiveResponse() {
if (shouldCloseConnection.get()) {
return;
}
//活动时间
touch();
try {
// See HBaseServer.Call.setResponse for where we write out the response.
// It writes the call.id (int), a flag byte, then optionally the length
// of the response (int) followed by data.
// Read the call id.
//请求的id
int id = in.readInt();
if (LOG.isDebugEnabled())
LOG.debug(getName() + " got value #" + id);
//从map中获取请求
Call call = calls.get(id);
//成功还是失败
// Read the flag byte
byte flag = in.readByte();
boolean isError = ResponseFlag.isError(flag);
if (ResponseFlag.isLength(flag)) {
// Currently length if present is unused.
in.readInt();
}
//RPC状态
int state = in.readInt(); // Read the state. Currently unused.
//如果请求异常，则获取异常信息，包装成异常
if (isError) {
if (call != null) {
//noinspection ThrowableInstanceNeverThrown
call.setException(new RemoteException(WritableUtils.readString(in),
WritableUtils.readString(in)));
}
} else {
//反序列化，HbaseObjectWritable，和server保持一致
Writable value = ReflectionUtils.newInstance(valueClass, conf);
value.readFields(in); // read value
// it's possible that this call may have been cleaned up due to a RPC
// timeout, so check if it still exists before setting the value.
//call还没被删掉，则设置返回值
if (call != null) {
call.setValue(value);
}
}
//完事后，从map中删除
calls.remove(id);
} catch (IOException e) {
if (e instanceof SocketTimeoutException && remoteId.rpcTimeout > 0) {
// Clean up open calls but don't treat this as a fatal condition,
// since we expect certain responses to not make it by the specified
// {@link ConnectionId#rpcTimeout}.
closeException = e;
} else {
// Since the server did not respond within the default ping interval
// time, treat this as a fatal condition and close this connection
markClosed(e);
}
} catch (Exception e) {
markClosed(new IOException(e.getMessage(), e));
} finally {
//检查超时请求并处理
if (remoteId.rpcTimeout > 0) {
cleanupCalls(remoteId.rpcTimeout);
}
}
}
超时请求检查和处理
protected void cleanupCalls(long rpcTimeout) {
Iterator<Entry<Integer, Call>> itor = calls.entrySet().iterator();
while (itor.hasNext()) {
Call c = itor.next().getValue();
long waitTime = System.currentTimeMillis() - c.getStartTime();
//超时了。写回一个超时异常
if (waitTime >= rpcTimeout) {
if (this.closeException == null) {
// There may be no exception in the case that there are many calls
// being multiplexed over this connection and these are succeeding
// fine while this Call object is taking a long time to finish
// over on the server; e.g. I just asked the regionserver to bulk
// open 3k regions or its a big fat multiput into a heavily-loaded
// server (Perhaps this only happens at the extremes?)
this.closeException = new CallTimeoutException("Call id=" + c.id +
", waitTime=" + waitTime + ", rpcTimetout=" + rpcTimeout);
}
c.setException(this.closeException);
synchronized (c) {
c.notifyAll();
}
itor.remove();
}
//calls是一个排序map，按照key的字典序，key是id，一个integer，也就是按照id升序，如果最老的请求(id最小)还没超时，则后续请求都没超时，直接break
else {
break;
}
}
//修改SO_TIMEOUT
try {
if (!calls.isEmpty()) {
Call firstCall = calls.get(calls.firstKey());
long maxWaitTime = System.currentTimeMillis() - firstCall.getStartTime();
if (maxWaitTime < rpcTimeout) {
rpcTimeout -= maxWaitTime;
}
}
if (!shouldCloseConnection.get()) {
closeException = null;
if (socket != null) {
socket.setSoTimeout((int) rpcTimeout);
}
}
} catch (SocketException e) {
LOG.debug("Couldn't lower timeout, which may result in longer than expected calls");
}
}
}
Reader/WRITER IO操作
int doIO(ByteBuffer buf, int ops) throws IOException {
/* For now only one thread is allowed. If user want to read or write
* from multiple threads, multiple streams could be created. In that
* case multiple threads work as well as underlying channel supports it.
*/
if (!buf.hasRemaining()) {
throw new IllegalArgumentException("Buffer has no data left.");
//or should we just return 0?
}
while (buf.hasRemaining()) {
if (closed) {
return -1;
}
//这里执行reader/writer的操作，基本就是对channel的read/write操作
try {
int n = performIO(buf);
//有数据处理，则返回，有可能只处理了部分数据，上层保证数据是否完整
if (n != 0) {
// successful io or an error.
return n;
}
} catch (IOException e) {
if (!channel.isOpen()) {
closed = true;
}
throw e;
}
//now wait for socket to be ready.
//一个数据都没处理，则继续监听对应READ/WRITE事件
int count = 0;
try {
count = selector.select(channel, ops, timeout);
} catch (IOException e) { //unexpected IOException.
closed = true;
throw e;
}
//超时了，否则继续执行读写操作，直到buffer被全部处理
if (count == 0) {
throw new SocketTimeoutException(timeoutExceptionString(channel,
timeout, ops));
}
// otherwise the socket should be ready for io.
}
return 0; // does not reach here.
}
IO read线程启动之后，业务线程开始写数据
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);
//占位
d.writeInt(0xdeadbeef); // placeholder for data length
//请求id，4个字节
d.writeInt(call.id);
//把param序列化
call.param.write(d);
//要写的数据
byte[] data = d.getData();
int dataLength = d.getLength();
// fill in the placeholder
//替换
Bytes.putInt(data, 0, dataLength - 4);
//noinspection SynchronizeOnNonFinalField
//写并flush
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);
}
}
SocketOutputStream写
public void write(byte[] b, int off, int len) throws IOException {
ByteBuffer buf = ByteBuffer.wrap(b, off, len);
//循环写，直到写完或抛异常
while (buf.hasRemaining()) {
try {
if (write(buf) < 0) {
throw new IOException("The stream is closed");
}
} catch (IOException e) {
/* Unlike read, write can not inform user of partial writes.
* So will close this if there was a partial write.
*/
if (buf.capacity() > buf.remaining()) {
writer.close();
}
throw e;
}
}
}
业务线程发送请求后，就进入等待状态，read线程则等待server端返回的数据，将返回数据反序列化后，放入call对象，并唤醒业务线程继续处理
相关推荐
在HBase的RPC框架中，我们可以看到它将通信的角色分配为客户端和服务端。客户端通过调用服务端提供的接口进行远程调用，而服务端则实现这些接口以响应客户端的请求。为了实现这一通信机制，HBase中定义了三个通信...
在HBase的架构中，Client是用户与系统交互的接口，它通过远程过程调用（RPC）机制与HMaster和HRegionServer通信。对于数据读写操作，Client直接与HRegionServer交互，而对于表管理和元数据操作，Client则与HMaster...
`hbase-protocol-1.2.6.jar`定义了HBase的RPC协议，用于客户端与服务器之间的通信。 为了处理Zookeeper，你需要`zookeeper-3.4.6.jar`，这是一个分布式协调服务，HBase使用它来管理集群的状态信息和元数据。`...
HBase则是建立在Hadoop之上的分布式列式数据库，特别适合实时查询大规模数据集。 标题“hadoop2.73-eclipse开发hbase所需要的所有jar包”表明这是一个针对Eclipse的开发资源包，包含了使用Hadoop 2.7.3版本开发...
HBase构建在Hadoop的HDFS之上，这意味着HBase利用HDFS来存储其底层数据，从而获得HDFS提供的高可靠性和高容错性。 4. **消息通信机制** HBase使用Apache Zookeeper来提供消息通信机制，这包括协调服务、命名服务...
- **hbase-protocol.jar**：提供了HBase的RPC协议，客户端和服务端通信需要。 - **zookeeper.jar**：HBase依赖Zookeeper进行集群协调，所以需要Zookeeper的JAR。 - **slf4j-api.jar**和**slf4j-log4j12.jar**：...
Hadoop是一个分布式存储和计算框架，而HBase是一个构建在Hadoop之上的非关系型数据库（NoSQL），特别适合处理大规模数据。这里我们将详细探讨Java如何与这两个组件进行交互，并重点关注所需的jar包。 首先，Java...
在异步与吞吐优化中，针对流式计算对实时性高和秒级毛刺的挑战，阿里集团通过基于Netty实现non-blocking client以及基于protobuf的non-blocking Stub/RpcCallback来提升吞吐。在与Flink集成后，实测吞吐较同步模式...
Client是整个HBase系统的入口客户端使用RPC协议与HMaster和RegionServer进行通信对于管理类（表的增删）操作，Client与HMaster进行RPC通信对于数据读写类操作Client与RegionServer进行RPC交互客户端可以是多个，也...
1. **Client**：客户端通过RPC（Remote Procedure Call）与HBase交互，既可以直接与HMaster通信进行管理操作，也可以直接与HRegionServer通信进行数据读写。 2. **Zookeeper**：Zookeeper集群在HBase中扮演着至关...
3. **hbase-protocol.jar**：包含了HBase的RPC协议，用于客户端和HBase服务器之间的通信。 4. **hbase-server.jar**：如果在客户端执行操作，如管理表或Region，可能需要这个服务器端的jar包，尽管通常情况下只需要...
4. **Hbase/Hadoop集成及应用开发**：HBase是一个构建在Hadoop之上的分布式列式存储系统，适用于存储海量结构化数据。与Hadoop结合可以实现高性能的数据存储和查询。 5. **YARN框架源码分析**：深入理解YARN的工作...
在Hadoop的RPC实现方法中，Client类、Server类、RPC类以及HDFS通信协议组是核心。这些组件共同协作实现远程过程调用，使得HDFS中的各个组件能够相互交流和协作。通过这些组件，HDFS能够处理客户端请求，并在集群内部...
- **Hadoop底层IPC原理和RPC**：探讨Hadoop内部通信机制，包括RPC（远程过程调用）的实现细节。 - **Hadoop底层googleProtoBuf的协议分析**：分析Google Protobuf在Hadoop中的应用情况。 #### 四、分布式数据库...
ZooKeeper为HBase提供了分布式协调服务，使得HBase能够实现一致性选举、故障恢复等功能。 12. **LSM数据结构** - **知识点**：Log-Structured Merge Tree (LSM Tree)的工作原理。 - **详细解析**：LSM是一种数据...