一。为何使用分布式锁?
当应用服务器数量超过1台,对相同数据的访问可能造成访问冲突(特别是写冲突)。单纯使用关系数据库比如MYSQL的应用可以借助于事务来实现锁,也可以使用版本号等实现乐观锁,最大的缺陷就是可用性降低(性能差)。对于GLEASY这种满足大规模并发访问请求的应用来说,使用数据库事务来实现数据库就有些捉襟见肘了。另外对于一些不依赖数据库的应用,比如分布式文件系统,为了保证同一文件在大量读写操作情况下的正确性,必须引入分布式锁来约束对同一文件的并发操作。
二。对分布式锁的要求
1.高性能(分布式锁不能成为系统的性能瓶颈)
2.避免死锁(拿到锁的结点挂掉不会导致其它结点永远无法继续)
3.支持锁重入
三。方案1,基于zookeeper的分布式锁
/**
* DistributedLockUtil.java
* 分布式锁工厂类,所有分布式请求都由该工厂类负责
**/
public class DistributedLockUtil {
private static Object schemeLock = new Object();
private static Object mutexLock = new Object();
private static Map<String,Object> mutexLockMap = new ConcurrentHashMap();
private String schema;
private Map<String,DistributedReentrantLock> cache = new ConcurrentHashMap<String,DistributedReentrantLock>();
private static Map<String,DistributedLockUtil> instances = new ConcurrentHashMap();
public static DistributedLockUtil getInstance(String schema){
DistributedLockUtil u = instances.get(schema);
if(u==null){
synchronized(schemeLock){
u = instances.get(schema);
if(u == null){
u = new DistributedLockUtil(schema);
instances.put(schema, u);
}
}
}
return u;
}
private DistributedLockUtil(String schema){
this.schema = schema;
}
private Object getMutex(String key){
Object mx = mutexLockMap.get(key);
if(mx == null){
synchronized(mutexLock){
mx = mutexLockMap.get(key);
if(mx==null){
mx = new Object();
mutexLockMap.put(key,mx);
}
}
}
return mx;
}
private DistributedReentrantLock getLock(String key){
DistributedReentrantLock lock = cache.get(key);
if(lock == null){
synchronized(getMutex(key)){
lock = cache.get(key);
if(lock == null){
lock = new DistributedReentrantLock(key,schema);
cache.put(key, lock);
}
}
}
return lock;
}
public void reset(){
for(String s : cache.keySet()){
getLock(s).unlock();
}
}
/**
* 尝试加锁
* 如果当前线程已经拥有该锁的话,直接返回false,表示不用再次加锁,此时不应该再调用unlock进行解锁
*
* @param key
* @return
* @throws InterruptedException
* @throws KeeperException
*/
public LockStat lock(String key) throws InterruptedException, KeeperException{
if(getLock(key).isOwner()){
return LockStat.NONEED;
}
getLock(key).lock();
return LockStat.SUCCESS;
}
public void clearLock(String key) throws InterruptedException, KeeperException{
synchronized(getMutex(key)){
DistributedReentrantLock l = cache.get(key);
l.clear();
cache.remove(key);
}
}
public void unlock(String key,LockStat stat) throws InterruptedException, KeeperException{
unlock(key,stat,false);
}
public void unlock(String key,LockStat stat,boolean keepalive) throws InterruptedException, KeeperException{
if(stat == null) return;
if(LockStat.SUCCESS.equals(stat)){
DistributedReentrantLock lock = getLock(key);
boolean hasWaiter = lock.unlock();
if(!hasWaiter && !keepalive){
synchronized(getMutex(key)){
lock.clear();
cache.remove(key);
}
}
}
}
public static enum LockStat{
NONEED,
SUCCESS
}
}
/**
*DistributedReentrantLock.java
*本地线程之间锁争用,先使用虚拟机内部锁机制,减少结点间通信开销
*/
public class DistributedReentrantLock {
private static final Logger logger = Logger.getLogger(DistributedReentrantLock.class);
private ReentrantLock reentrantLock = new ReentrantLock();
private WriteLock writeLock;
private long timeout = 3*60*1000;
private final Object mutex = new Object();
private String dir;
private String schema;
private final ExitListener exitListener = new ExitListener(){
@Override
public void execute() {
initWriteLock();
}
};
private synchronized void initWriteLock(){
logger.debug("初始化writeLock");
writeLock = new WriteLock(dir,new LockListener(){
@Override
public void lockAcquired() {
synchronized(mutex){
mutex.notify();
}
}
@Override
public void lockReleased() {
}
},schema);
if(writeLock != null && writeLock.zk != null){
writeLock.zk.addExitListener(exitListener);
}
synchronized(mutex){
mutex.notify();
}
}
public DistributedReentrantLock(String dir,String schema) {
this.dir = dir;
this.schema = schema;
initWriteLock();
}
public void lock(long timeout) throws InterruptedException, KeeperException {
reentrantLock.lock();//多线程竞争时,先拿到第一层锁
try{
boolean res = writeLock.trylock();
if(!res){
synchronized(mutex){
mutex.wait(timeout);
}
if(writeLock == null || !writeLock.isOwner()){
throw new InterruptedException("锁超时");
}
}
}catch(InterruptedException e){
reentrantLock.unlock();
throw e;
}catch(KeeperException e){
reentrantLock.unlock();
throw e;
}
}
public void lock() throws InterruptedException, KeeperException {
lock(timeout);
}
public void destroy() throws KeeperException {
writeLock.unlock();
}
public boolean unlock(){
if(!isOwner()) return false;
try{
writeLock.unlock();
reentrantLock.unlock();//多线程竞争时,释放最外层锁
}catch(RuntimeException e){
reentrantLock.unlock();//多线程竞争时,释放最外层锁
throw e;
}
return reentrantLock.hasQueuedThreads();
}
public boolean isOwner() {
return reentrantLock.isHeldByCurrentThread() && writeLock.isOwner();
}
public void clear() {
writeLock.clear();
}
}
/**
*WriteLock.java
*基于zk的锁实现
*一个最简单的场景如下:
*1.结点A请求加锁,在特定路径下注册自己(会话自增结点),得到一个ID号1
*2.结点B请求加锁,在特定路径下注册自己(会话自增结点),得到一个ID号2
*3.结点A获取所有结点ID,判断出来自己是最小结点号,于是获得锁
*4.结点B获取所有结点ID,判断出来自己不是最小结点,于是监听小于自己的最大结点(结点A)变更事件
*5.结点A拿到锁,处理业务,处理完,释放锁(删除自己)
*6.结点B收到结点A变更事件,判断出来自己已经是最小结点号,于是获得锁。
*/
public class WriteLock extends ZkPrimative {
private static final Logger LOG = Logger.getLogger(WriteLock.class);
private final String dir;
private String id;
private LockNode idName;
private String ownerId;
private String lastChildId;
private byte[] data = {0x12, 0x34};
private LockListener callback;
public WriteLock(String dir,String schema) {
super(schema,true);
this.dir = dir;
}
public WriteLock(String dir,LockListener callback,String schema) {
this(dir,schema);
this.callback = callback;
}
public LockListener getLockListener() {
return this.callback;
}
public void setLockListener(LockListener callback) {
this.callback = callback;
}
public synchronized void unlock() throws RuntimeException {
if(zk == null || zk.isClosed()){
return;
}
if (id != null) {
try {
zk.delete(id, -1);
} catch (InterruptedException e) {
LOG.warn("Caught: " + e, e);
//set that we have been interrupted.
Thread.currentThread().interrupt();
} catch (KeeperException.NoNodeException e) {
// do nothing
} catch (KeeperException e) {
LOG.warn("Caught: " + e, e);
throw (RuntimeException) new RuntimeException(e.getMessage()).
initCause(e);
}finally {
if (callback != null) {
callback.lockReleased();
}
id = null;
}
}
}
private class LockWatcher implements Watcher {
public void process(WatchedEvent event) {
LOG.debug("Watcher fired on path: " + event.getPath() + " state: " +
event.getState() + " type " + event.getType());
try {
trylock();
} catch (Exception e) {
LOG.warn("Failed to acquire lock: " + e, e);
}
}
}
private void findPrefixInChildren(String prefix, ZooKeeper zookeeper, String dir)
throws KeeperException, InterruptedException {
List<String> names = zookeeper.getChildren(dir, false);
for (String name : names) {
if (name.startsWith(prefix)) {
id = dir + "/" + name;
if (LOG.isDebugEnabled()) {
LOG.debug("Found id created last time: " + id);
}
break;
}
}
if (id == null) {
id = zookeeper.create(dir + "/" + prefix, data,
acl, EPHEMERAL_SEQUENTIAL);
if (LOG.isDebugEnabled()) {
LOG.debug("Created id: " + id);
}
}
}
public void clear() {
if(zk == null || zk.isClosed()){
return;
}
try {
zk.delete(dir, -1);
} catch (Exception e) {
LOG.error("clear error: " + e,e);
}
}
public synchronized boolean trylock() throws KeeperException, InterruptedException {
if(zk == null){
LOG.info("zk 是空");
return false;
}
if (zk.isClosed()) {
LOG.info("zk 已经关闭");
return false;
}
ensurePathExists(dir);
LOG.debug("id:"+id);
do {
if (id == null) {
long sessionId = zk.getSessionId();
String prefix = "x-" + sessionId + "-";
idName = new LockNode(id);
LOG.debug("idName:"+idName);
}
if (id != null) {
List<String> names = zk.getChildren(dir, false);
if (names.isEmpty()) {
LOG.warn("No children in: " + dir + " when we've just " +
"created one! Lets recreate it...");
id = null;
} else {
SortedSet<LockNode> sortedNames = new TreeSet<LockNode>();
for (String name : names) {
sortedNames.add(new LockNode(dir + "/" + name));
}
ownerId = sortedNames.first().getName();
LOG.debug("all:"+sortedNames);
SortedSet<LockNode> lessThanMe = sortedNames.headSet(idName);
LOG.debug("less than me:"+lessThanMe);
if (!lessThanMe.isEmpty()) {
LockNode lastChildName = lessThanMe.last();
lastChildId = lastChildName.getName();
if (LOG.isDebugEnabled()) {
LOG.debug("watching less than me node: " + lastChildId);
}
Stat stat = zk.exists(lastChildId, new LockWatcher());
if (stat != null) {
return Boolean.FALSE;
} else {
LOG.warn("Could not find the" +
" stats for less than me: " + lastChildName.getName());
}
} else {
if (isOwner()) {
if (callback != null) {
callback.lockAcquired();
}
return Boolean.TRUE;
}
}
}
}
}
while (id == null);
return Boolean.FALSE;
}
public String getDir() {
return dir;
}
public boolean isOwner() {
return id != null && ownerId != null && id.equals(ownerId);
}
public String getId() {
return this.id;
}
}
使用本方案实现的分布式锁,可以很好地解决锁重入的问题,而且使用会话结点来避免死锁;性能方面,根据笔者自测结果,加锁解锁各一次算是一个操作,本方案实现的分布式锁,TPS大概为2000-3000,性能比较一般;