前言
本篇文章介绍zookeeper watcher工作机制，是上一篇zookeeper新建节点请求源码的延续。
客户端
客户端通过zookeeper.addWatch方法设置对指定的节点的事件监听
public void addWatch(String basePath, Watcher watcher, AddWatchMode mode)
throws KeeperException, InterruptedException {
PathUtils.validatePath(basePath);
//根据chroot和用户指定basePath生成服务端serverPath
String serverPath = prependChroot(basePath);
//下面是创建对指定节点设置watcher监控对象的过程
RequestHeader h = new RequestHeader();
h.setType(ZooDefs.OpCode.addWatch);
AddWatchRequest request = new AddWatchRequest(serverPath, mode.getMode());
//把请求通过连接发送到服务端
ReplyHeader r = cnxn.submitRequest(h, request, new ErrorResponse(),
new AddWatchRegistration(watcher, basePath, mode));
if (r.getErr() != 0) {
throw KeeperException.create(KeeperException.Code.get(r.getErr()),
basePath);
}
}
服务端
服务端接受到的请求也会像在zookeeper新建节点请求源码描述的那样经历如下几个处理流程：
process.png
所有的流程过程都是相同的，不同的地方就是那些case:xxx的地方，对于watch来说不同的地方就是FinalRequestProcessor.processRequest中的下面这段代码
case OpCode.addWatch: {
lastOp = "ADDW";
AddWatchRequest addWatcherRequest = new AddWatchRequest();
ByteBufferInputStream.byteBuffer2Record(request.request,
addWatcherRequest);
zks.getZKDatabase().addWatch(addWatcherRequest.getPath(), cnxn, addWatcherRequest.getMode());
rsp = new ErrorResponse(0);
break;
}
我们可以看到当请求是watcher类型的时候，会把请交给ZKDatabase.addWatch-->DataTree.addWatch来处理，我们看下DataTree.addWatch的源代码
public void addWatch(String basePath, Watcher watcher, int mode) {
//获取watcherMode（有三种watcherMode，标准类型，持久化类型，持久化循环型）
WatcherMode watcherMode = WatcherMode.fromZooDef(mode);
dataWatches.addWatch(basePath, watcher, watcherMode);
childWatches.addWatch(basePath, watcher, watcherMode);
}
上面我们介绍了三种类型的watcher，现在我们逐一解释下他们的作用
- 标准型: 这种类型的watcher在被触发之后就会消失，如果想继续监听对应的节点的事件，那么需要重新注册
- 持久化型： 这种类型watcher在被触发之后不会消失，后面如果有对应节点事件发生，watcher会被持续触发
- 持久化循环型：如果一个节点设置了这种类型的watcher，那么如果这个节点的子节点发生了变化，那么这个节点的上层节点（父节点，父节点的父节点........）也会被触发该事件
WatcherManager.addWatch
服务端的监控管理器添加watcher的过程如下
public synchronized boolean addWatch(String path, Watcher watcher, WatcherMode watcherMode) {
//对于服务端来说节点路径对应的Watcher是NioServerCnxn
if (isDeadWatcher(watcher)) {
//如果客户端和服务端断开了连接，那么服务端对应的连接对象NioServerCnxn就会被设置为dead
LOG.debug("Ignoring addWatch with closed cnxn");
return false;
}
//watchTable存储了节点路径对应的Watcher列表
Set list = watchTable.get(path);
if (list == null) {
// don't waste memory if there are few watches on a node
// rehash when the 4th entry is added, doubling size thereafter
// seems like a good compromise
//设置存放watcher的列表，默认初始容量是4
list = new HashSet<>(4);
watchTable.put(path, list);
}
list.add(watcher);
//watch2Paths存放的是watcher和路径的关系
Set paths = watch2Paths.get(watcher);
if (paths == null) {
// cnxns typically have many watches, so use default cap here
paths = new HashSet<>();
watch2Paths.put(watcher, paths);
}
watcherModeManager.setWatcherMode(watcher, path, watcherMode);
return paths.add(path);
}
通过上面的代码我们可以发现路径对应的watcher被加入到了dataWatches和childWatches中，这样我们结合在zookeeper新建节点请求源码最后事件触发的部分，继续去分析事件是如何被触发
triggerWatch
最终WatchManager.triggerWatch会被调用
public WatcherOrBitSet triggerWatch(String path, EventType type, WatcherOrBitSet supress) {
WatchedEvent e = new WatchedEvent(type, KeeperState.SyncConnected, path);
Set watchers = new HashSet<>();
//这个获取节点的代表对象PathParentIterator（如果watch是循环型的，那么节点的上层节点也会被触发）
PathParentIterator pathParentIterator = getPathParentIterator(path);
synchronized (this) {
for (String localPath : pathParentIterator.asIterable()) {
//根据节点路径获取对应的watcher
Set thisWatchers = watchTable.get(localPath);
if (thisWatchers == null || thisWatchers.isEmpty()) {
continue;
}
Iterator iterator = thisWatchers.iterator();
while (iterator.hasNext()) {
//遍历每一个watcher
Watcher watcher = iterator.next();
WatcherMode watcherMode = watcherModeManager.getWatcherMode(watcher, localPath);
if (watcherMode.isRecursive()) {
if (type != EventType.NodeChildrenChanged) {
//对于是循环型的watcher，如果发生的是NodeChildrenChanged的事件，那么把被watcher加入到被触发的列表中
watchers.add(watcher);
}
} else if (!pathParentIterator.atParentPath()) {
//对于非循环类型的节点，不会触发父节点的watcher，只有是路径对应节点的watcher会被触发
watchers.add(watcher);
if (!watcherMode.isPersistent()) {
//如果watcher不是持久化的类型，那么把它删除，对于这种类型的watcher每次触发之后，需要重新注册才能继续监听事件
iterator.remove();
Set paths = watch2Paths.get(watcher);
if (paths != null) {
paths.remove(localPath);
}
}
}
}
if (thisWatchers.isEmpty()) {
//如果path对应的watcher为空，那么把path从watchTable中删除
watchTable.remove(localPath);
}
}
}
if (watchers.isEmpty()) {
if (LOG.isTraceEnabled()) {
ZooTrace.logTraceMessage(LOG, ZooTrace.EVENT_DELIVERY_TRACE_MASK, "No watchers for " + path);
}
return null;
}
//下面是对找到了的watcher一个一个执行process方法来处理每个事件
for (Watcher w : watchers) {
if (supress != null && supress.contains(w)) {
continue;
}
w.process(e);
}
switch (type) {
case NodeCreated:
ServerMetrics.getMetrics().NODE_CREATED_WATCHER.add(watchers.size());
break;
case NodeDeleted:
ServerMetrics.getMetrics().NODE_DELETED_WATCHER.add(watchers.size());
break;
case NodeDataChanged:
ServerMetrics.getMetrics().NODE_CHANGED_WATCHER.add(watchers.size());
break;
case NodeChildrenChanged:
ServerMetrics.getMetrics().NODE_CHILDREN_WATCHER.add(watchers.size());
break;
default:
// Other types not logged.
break;
}
return new WatcherOrBitSet(watchers);
}
NIOServerCnxn.process
接下来我们看下watcher.process,对于服务端来说watcher的实例是NIOServerCnxn，我们先看下NIOServerCnxn.process的源代码
public void process(WatchedEvent event) {
//生成返回头信息
ReplyHeader h = new ReplyHeader(ClientCnxn.NOTIFICATION_XID, -1L, 0);
if (LOG.isTraceEnabled()) {
ZooTrace.logTraceMessage(
LOG,
ZooTrace.EVENT_DELIVERY_TRACE_MASK,
"Deliver event " + event + " to 0x" + Long.toHexString(this.sessionId) + " through " + this);
}
// Convert WatchedEvent to a type that can be sent over the wire
WatcherEvent e = event.getWrapper();
// The last parameter OpCode here is used to select the response cache.
// Passing OpCode.error (with a value of -1) means we don't care, as we don't need
// response cache on delivering watcher events.
//把返回信息发送给客户端
sendResponse(h, e, "notification", null, null, ZooDefs.OpCode.error);
}
客户端处理监控response
ClientCnxn.readResponse
readResponse是客户端处理服务端响应数据的地方
void readResponse(ByteBuffer incomingBuffer) throws IOException {
ByteBufferInputStream bbis = new ByteBufferInputStream(incomingBuffer);
BinaryInputArchive bbia = BinaryInputArchive.getArchive(bbis);
ReplyHeader replyHdr = new ReplyHeader();
//反序列化服务端返回的响应对象头
replyHdr.deserialize(bbia, "header");
//下面是根据响应头的类型做不同的处理
switch (replyHdr.getXid()) {
case PING_XID:
//这里是处理客户端发送的心跳检测response的处理逻辑
LOG.debug("Got ping response for session id: 0x{} after {}ms.",
Long.toHexString(sessionId),
((System.nanoTime() - lastPingSentNs) / 1000000));
return;
case AUTHPACKET_XID:
//这里是客户端处理权限验证response的逻辑
LOG.debug("Got auth session id: 0x{}", Long.toHexString(sessionId));
if (replyHdr.getErr() == KeeperException.Code.AUTHFAILED.intValue()) {
state = States.AUTH_FAILED;
eventThread.queueEvent(new WatchedEvent(Watcher.Event.EventType.None,
Watcher.Event.KeeperState.AuthFailed, null));
eventThread.queueEventOfDeath();
}
return;
case NOTIFICATION_XID:
//这里是客户端处理添加节点监听response
LOG.debug("Got notification session id: 0x{}",
Long.toHexString(sessionId));
WatcherEvent event = new WatcherEvent();
//反序列化监听事件response
event.deserialize(bbia, "response");
// convert from a server path to a client path
if (chrootPath != null) {
String serverPath = event.getPath();
if (serverPath.compareTo(chrootPath) == 0) {
event.setPath("/");
} else if (serverPath.length() > chrootPath.length()) {
event.setPath(serverPath.substring(chrootPath.length()));
} else {
LOG.warn("Got server path {} which is too short for chroot path {}.",
event.getPath(), chrootPath);
}
}
//根据WatchEvent生成WatchedEvent
WatchedEvent we = new WatchedEvent(event);
LOG.debug("Got {} for session id 0x{}", we, Long.toHexString(sessionId));
//事件最终交给eventThread处理
eventThread.queueEvent(we);
return;
default:
break;
}
// If SASL authentication is currently in progress, construct and
// send a response packet immediately, rather than queuing a
// response as with other packets.
if (tunnelAuthInProgress()) {
GetSASLRequest request = new GetSASLRequest();
request.deserialize(bbia, "token");
zooKeeperSaslClient.respondToServer(request.getToken(), ClientCnxn.this);
return;
}
Packet packet;
//如果是别的类型的事物请求，那么从pendingQueue中取出第一个请求，
//因为客户端的请求都会被顺序的发送和顺序的处理，
// 所以取出的这个请求是最早被发送并且还没有得到服务端返回的请求
synchronized (pendingQueue) {
if (pendingQueue.size() == 0) {
throw new IOException("Nothing in the queue, but got " + replyHdr.getXid());
}
packet = pendingQueue.remove();
}
/*
* Since requests are processed in order, we better get a response
* to the first request!
*/
try {
//比较请求头的xid和响应头的xid，理论上他们应该相等，如果不相等说明客户端事物顺序已经破坏，那么需要返回异常信息
if (packet.requestHeader.getXid() != replyHdr.getXid()) {
packet.replyHeader.setErr(KeeperException.Code.CONNECTIONLOSS.intValue());
throw new IOException("Xid out of order. Got Xid " + replyHdr.getXid()
+ " with err " + replyHdr.getErr()
+ " expected Xid " + packet.requestHeader.getXid()
+ " for a packet with details: " + packet);
}
packet.replyHeader.setXid(replyHdr.getXid());
packet.replyHeader.setErr(replyHdr.getErr());
packet.replyHeader.setZxid(replyHdr.getZxid());
if (replyHdr.getZxid() > 0) {
//设置客户端最新处理的事物id
lastZxid = replyHdr.getZxid();
}
if (packet.response != null && replyHdr.getErr() == 0) {
//反序列化服务端的处理结果
packet.response.deserialize(bbia, "response");
}
LOG.debug("Reading reply session id: 0x{}, packet:: {}", Long.toHexString(sessionId), packet);
} finally {
//最后一步处理：这里面会设置响应头状态，调用回调方法，调用eventThread按照watcher的不同注册watcher到不同的监听容器中，等待将来被调用等等其他一系列操作
finishPacket(packet);
}
}
eventThread.queueEvent
queueEvent把事件加入到客户端事件队列中
private void queueEvent(WatchedEvent event, Set materializedWatchers) {
if (event.getType() == EventType.None && sessionState == event.getState()) {
return;
}
sessionState = event.getState();
final Set watchers;
if (materializedWatchers == null) {
// materialize the watchers based on the event
//从根据事件类型和事件发生的节点路径，把客户端用户自定义的watcher从各种watcher容器中取出来等待被触发
watchers = watcher.materialize(event.getState(), event.getType(), event.getPath());
} else {
watchers = new HashSet();
watchers.addAll(materializedWatchers);
}
//形成WatcherSetEventPair对象然后加入waitingEvents队列中
WatcherSetEventPair pair = new WatcherSetEventPair(watchers, event);
// queue the pair (watch set & event) for later processing
waitingEvents.add(pair);
}
接下来我们看下eventThread的处理逻辑,我们先看eventThread的run方法
public void run() {
try {
isRunning = true;
while (true) {
//从waitingEvents队列中取出一个事件
Object event = waitingEvents.take();
if (event == eventOfDeath) {
wasKilled = true;
} else {
//处理事件
processEvent(event);
}
if (wasKilled) {
synchronized (waitingEvents) {
if (waitingEvents.isEmpty()) {
isRunning = false;
break;
}
}
}
}
} catch (InterruptedException e) {
LOG.error("Event thread exiting due to interruption", e);
}
LOG.info("EventThread shut down for session: 0x{}", Long.toHexString(getSessionId()));
}
processEvent
private void processEvent(Object event) {
try {
if (event instanceof WatcherSetEventPair) {
// each watcher will process the event
WatcherSetEventPair pair = (WatcherSetEventPair) event;
for (Watcher watcher : pair.watchers) {
try {
//调用用户自己定义的watcher.process方法
watcher.process(pair.event);
} catch (Throwable t) {
LOG.error("Error while calling watcher ", t);
}
}
} else if (event instanceof LocalCallback) {
//下面是对callback的处理逻辑，这里忽略不讲解了
LocalCallback lcb = (LocalCallback) event;
if (lcb.cb instanceof StatCallback) {
((StatCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null);
} else if (lcb.cb instanceof DataCallback) {
((DataCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null, null);
} else if (lcb.cb instanceof ACLCallback) {
((ACLCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null, null);
} else if (lcb.cb instanceof ChildrenCallback) {
((ChildrenCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null);
} else if (lcb.cb instanceof Children2Callback) {
((Children2Callback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null, null);
} else if (lcb.cb instanceof StringCallback) {
((StringCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, null);
} else if (lcb.cb instanceof AsyncCallback.EphemeralsCallback) {
((AsyncCallback.EphemeralsCallback) lcb.cb).processResult(lcb.rc, lcb.ctx, null);
} else if (lcb.cb instanceof AsyncCallback.AllChildrenNumberCallback) {
((AsyncCallback.AllChildrenNumberCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx, -1);
} else {
((VoidCallback) lcb.cb).processResult(lcb.rc, lcb.path, lcb.ctx);
}
} else {
//下面是对packet类型的处理
Packet p = (Packet) event;
int rc = 0;
String clientPath = p.clientPath;
if (p.replyHeader.getErr() != 0) {
rc = p.replyHeader.getErr();
}
if (p.cb == null) {
LOG.warn("Somehow a null cb got to EventThread!");
} else if (p.response instanceof ExistsResponse
|| p.response instanceof SetDataResponse
|| p.response instanceof SetACLResponse) {
StatCallback cb = (StatCallback) p.cb;
if (rc == 0) {
if (p.response instanceof ExistsResponse) {
cb.processResult(rc, clientPath, p.ctx, ((ExistsResponse) p.response).getStat());
} else if (p.response instanceof SetDataResponse) {
cb.processResult(rc, clientPath, p.ctx, ((SetDataResponse) p.response).getStat());
} else if (p.response instanceof SetACLResponse) {
cb.processResult(rc, clientPath, p.ctx, ((SetACLResponse) p.response).getStat());
}
} else {
cb.processResult(rc, clientPath, p.ctx, null);
}
} else if (p.response instanceof GetDataResponse) {
DataCallback cb = (DataCallback) p.cb;
GetDataResponse rsp = (GetDataResponse) p.response;
if (rc == 0) {
cb.processResult(rc, clientPath, p.ctx, rsp.getData(), rsp.getStat());
} else {
cb.processResult(rc, clientPath, p.ctx, null, null);
}
} else if (p.response instanceof GetACLResponse) {
ACLCallback cb = (ACLCallback) p.cb;
GetACLResponse rsp = (GetACLResponse) p.response;
if (rc == 0) {
cb.processResult(rc, clientPath, p.ctx, rsp.getAcl(), rsp.getStat());
} else {
cb.processResult(rc, clientPath, p.ctx, null, null);
}
} else if (p.response instanceof GetChildrenResponse) {
ChildrenCallback cb = (ChildrenCallback) p.cb;
GetChildrenResponse rsp = (GetChildrenResponse) p.response;
if (rc == 0) {
cb.processResult(rc, clientPath, p.ctx, rsp.getChildren());
} else {
cb.processResult(rc, clientPath, p.ctx, null);
}
} else if (p.response instanceof GetAllChildrenNumberResponse) {
AllChildrenNumberCallback cb = (AllChildrenNumberCallback) p.cb;
GetAllChildrenNumberResponse rsp = (GetAllChildrenNumberResponse) p.response;
if (rc == 0) {
cb.processResult(rc, clientPath, p.ctx, rsp.getTotalNumber());
} else {
cb.processResult(rc, clientPath, p.ctx, -1);
}
} else if (p.response instanceof GetChildren2Response) {
Children2Callback cb = (Children2Callback) p.cb;
GetChildren2Response rsp = (GetChildren2Response) p.response;
if (rc == 0) {
cb.processResult(rc, clientPath, p.ctx, rsp.getChildren(), rsp.getStat());
} else {
cb.processResult(rc, clientPath, p.ctx, null, null);
}
} else if (p.response instanceof CreateResponse) {
StringCallback cb = (StringCallback) p.cb;
CreateResponse rsp = (CreateResponse) p.response;
if (rc == 0) {
cb.processResult(
rc,
clientPath,
p.ctx,
(chrootPath == null
? rsp.getPath()
: rsp.getPath().substring(chrootPath.length())));
} else {
cb.processResult(rc, clientPath, p.ctx, null);
}
} else if (p.response instanceof Create2Response) {
Create2Callback cb = (Create2Callback) p.cb;
Create2Response rsp = (Create2Response) p.response;
if (rc == 0) {
cb.processResult(
rc,
clientPath,
p.ctx,
(chrootPath == null
? rsp.getPath()
: rsp.getPath().substring(chrootPath.length())),
rsp.getStat());
} else {
cb.processResult(rc, clientPath, p.ctx, null, null);
}
} else if (p.response instanceof MultiResponse) {
MultiCallback cb = (MultiCallback) p.cb;
MultiResponse rsp = (MultiResponse) p.response;
if (rc == 0) {
List results = rsp.getResultList();
int newRc = rc;
for (OpResult result : results) {
if (result instanceof ErrorResult
&& KeeperException.Code.OK.intValue()
!= (newRc = ((ErrorResult) result).getErr())) {
break;
}
}
cb.processResult(newRc, clientPath, p.ctx, results);
} else {
cb.processResult(rc, clientPath, p.ctx, null);
}
} else if (p.response instanceof GetEphemeralsResponse) {
EphemeralsCallback cb = (EphemeralsCallback) p.cb;
GetEphemeralsResponse rsp = (GetEphemeralsResponse) p.response;
if (rc == 0) {
cb.processResult(rc, p.ctx, rsp.getEphemerals());
} else {
cb.processResult(rc, p.ctx, null);
}
} else if (p.cb instanceof VoidCallback) {
VoidCallback cb = (VoidCallback) p.cb;
cb.processResult(rc, clientPath, p.ctx);
}
}
} catch (Throwable t) {
LOG.error("Unexpected throwable", t);
}
}
}
上面就是客户度和服务端对watcher的处理逻辑，感谢耐心阅读