序
本文主要研究一下AsyncHttpClient的ChannelPool
ChannelPool
org/asynchttpclient/channel/ChannelPool.java
public interface ChannelPool {
/**
* Add a channel to the pool
*
* @param channel an I/O channel
* @param partitionKey a key used to retrieve the cached channel
* @return true if added.
*/
boolean offer(Channel channel, Object partitionKey);
/**
* Remove the channel associated with the uri.
*
* @param partitionKey the partition used when invoking offer
* @return the channel associated with the uri
*/
Channel poll(Object partitionKey);
/**
* Remove all channels from the cache. A channel might have been associated
* with several uri.
*
* @param channel a channel
* @return the true if the channel has been removed
*/
boolean removeAll(Channel channel);
/**
* Return true if a channel can be cached. A implementation can decide based
* on some rules to allow caching Calling this method is equivalent of
* checking the returned value of {@link ChannelPool#offer(Channel, Object)}
*
* @return true if a channel can be cached.
*/
boolean isOpen();
/**
* Destroy all channels that has been cached by this instance.
*/
void destroy();
/**
* Flush partitions based on a predicate
*
* @param predicate the predicate
*/
void flushPartitions(Predicate<Object> predicate);
/**
* @return The number of idle channels per host.
*/
Map<String, Long> getIdleChannelCountPerHost();
}
ChannelPool定义了offer、poll、removeAll、isOpen、destroy、flushPartitions、getIdleChannelCountPerHost方法,它有两个实现类,分别是NoopChannelPool及DefaultChannelPool
NoopChannelPool
org/asynchttpclient/channel/NoopChannelPool.java
public enum NoopChannelPool implements ChannelPool {
INSTANCE;
@Override
public boolean offer(Channel channel, Object partitionKey) {
return false;
}
@Override
public Channel poll(Object partitionKey) {
return null;
}
@Override
public boolean removeAll(Channel channel) {
return false;
}
@Override
public boolean isOpen() {
return true;
}
@Override
public void destroy() {
}
@Override
public void flushPartitions(Predicate<Object> predicate) {
}
@Override
public Map<String, Long> getIdleChannelCountPerHost() {
return Collections.emptyMap();
}
}
NoopChannelPool是个枚举,用枚举实现了单例,其方法默认为空操作
DefaultChannelPool
/**
* A simple implementation of {@link ChannelPool} based on a {@link java.util.concurrent.ConcurrentHashMap}
*/
public final class DefaultChannelPool implements ChannelPool {
private static final Logger LOGGER = LoggerFactory.getLogger(DefaultChannelPool.class);
private final ConcurrentHashMap<Object, ConcurrentLinkedDeque<IdleChannel>> partitions = new ConcurrentHashMap<>();
private final ConcurrentHashMap<ChannelId, ChannelCreation> channelId2Creation;
private final AtomicBoolean isClosed = new AtomicBoolean(false);
private final Timer nettyTimer;
private final int connectionTtl;
private final boolean connectionTtlEnabled;
private final int maxIdleTime;
private final boolean maxIdleTimeEnabled;
private final long cleanerPeriod;
private final PoolLeaseStrategy poolLeaseStrategy;
public DefaultChannelPool(AsyncHttpClientConfig config, Timer hashedWheelTimer) {
this(config.getPooledConnectionIdleTimeout(),
config.getConnectionTtl(),
hashedWheelTimer,
config.getConnectionPoolCleanerPeriod());
}
public DefaultChannelPool(int maxIdleTime,
int connectionTtl,
Timer nettyTimer,
int cleanerPeriod) {
this(maxIdleTime,
connectionTtl,
PoolLeaseStrategy.LIFO,
nettyTimer,
cleanerPeriod);
}
public DefaultChannelPool(int maxIdleTime,
int connectionTtl,
PoolLeaseStrategy poolLeaseStrategy,
Timer nettyTimer,
int cleanerPeriod) {
this.maxIdleTime = maxIdleTime;
this.connectionTtl = connectionTtl;
connectionTtlEnabled = connectionTtl > 0;
channelId2Creation = connectionTtlEnabled ? new ConcurrentHashMap<>() : null;
this.nettyTimer = nettyTimer;
maxIdleTimeEnabled = maxIdleTime > 0;
this.poolLeaseStrategy = poolLeaseStrategy;
this.cleanerPeriod = Math.min(cleanerPeriod, Math.min(connectionTtlEnabled ? connectionTtl : Integer.MAX_VALUE, maxIdleTimeEnabled ? maxIdleTime : Integer.MAX_VALUE));
if (connectionTtlEnabled || maxIdleTimeEnabled)
scheduleNewIdleChannelDetector(new IdleChannelDetector());
}
//......
}
DefaultChannelPool基于ConcurrentHashMap实现了ChannelPool接口,主要的参数为connectionTtl、maxIdleTime、cleanerPeriod、poolLeaseStrategy;cleanerPeriod会取connectionTtl、maxIdleTime、传入的cleanerPeriod的最小值;开启connectionTtl或者maxIdleTime的话,会往nettyTimer添加IdleChannelDetector,延后cleanerPeriod时间执行
offer
public boolean offer(Channel channel, Object partitionKey) {
if (isClosed.get())
return false;
long now = unpreciseMillisTime();
if (isTtlExpired(channel, now))
return false;
boolean offered = offer0(channel, partitionKey, now);
if (connectionTtlEnabled && offered) {
registerChannelCreation(channel, partitionKey, now);
}
return offered;
}
private boolean isTtlExpired(Channel channel, long now) {
if (!connectionTtlEnabled)
return false;
ChannelCreation creation = channelId2Creation.get(channel.id());
return creation != null && now - creation.creationTime >= connectionTtl;
}
private boolean offer0(Channel channel, Object partitionKey, long now) {
ConcurrentLinkedDeque<IdleChannel> partition = partitions.get(partitionKey);
if (partition == null) {
partition = partitions.computeIfAbsent(partitionKey, pk -> new ConcurrentLinkedDeque<>());
}
return partition.offerFirst(new IdleChannel(channel, now));
}
private void registerChannelCreation(Channel channel, Object partitionKey, long now) {
ChannelId id = channel.id();
if (!channelId2Creation.containsKey(id)) {
channelId2Creation.putIfAbsent(id, new ChannelCreation(now, partitionKey));
}
}
offer接口先判断isTtlExpired,如果channel的存活时间超过connectionTtl则返回false,否则执行offer0,往ConcurrentLinkedDeque<IdleChannel>添加,若添加成功且connectionTtlEnabled则执行registerChannelCreation,维护创建时间
poll
/**
* {@inheritDoc}
*/
public Channel poll(Object partitionKey) {
IdleChannel idleChannel = null;
ConcurrentLinkedDeque<IdleChannel> partition = partitions.get(partitionKey);
if (partition != null) {
while (idleChannel == null) {
idleChannel = poolLeaseStrategy.lease(partition);
if (idleChannel == null)
// pool is empty
break;
else if (!Channels.isChannelActive(idleChannel.channel)) {
idleChannel = null;
LOGGER.trace("Channel is inactive, probably remotely closed!");
} else if (!idleChannel.takeOwnership()) {
idleChannel = null;
LOGGER.trace("Couldn't take ownership of channel, probably in the process of being expired!");
}
}
}
return idleChannel != null ? idleChannel.channel : null;
}
poll方法是根据partitionKey找到对应的ConcurrentLinkedDeque<IdleChannel>,然后循环执行poolLeaseStrategy.lease(partition),若idleChannel为null直接break,若isChannelActive为false则重置为null继续循环,若idleChannel.takeOwnership()为false也重置为null继续循环
removeAll
/**
* {@inheritDoc}
*/
public boolean removeAll(Channel channel) {
ChannelCreation creation = connectionTtlEnabled ? channelId2Creation.remove(channel.id()) : null;
return !isClosed.get() && creation != null && partitions.get(creation.partitionKey).remove(new IdleChannel(channel, Long.MIN_VALUE));
}
removeAll方法会将指定的channel从channelId2Creation及ConcurrentLinkedDeque<IdleChannel>中移除
isOpen
/**
* {@inheritDoc}
*/
public boolean isOpen() {
return !isClosed.get();
}
isOpen则取的isClosed变量
destroy
/**
* {@inheritDoc}
*/
public void destroy() {
if (isClosed.getAndSet(true))
return;
partitions.clear();
if (connectionTtlEnabled) {
channelId2Creation.clear();
}
}
destroy会设置isClosed为true,然后清空partitions及channelId2Creation
flushPartitions
public void flushPartitions(Predicate<Object> predicate) {
for (Map.Entry<Object, ConcurrentLinkedDeque<IdleChannel>> partitionsEntry : partitions.entrySet()) {
Object partitionKey = partitionsEntry.getKey();
if (predicate.test(partitionKey))
flushPartition(partitionKey, partitionsEntry.getValue());
}
}
private void flushPartition(Object partitionKey, ConcurrentLinkedDeque<IdleChannel> partition) {
if (partition != null) {
partitions.remove(partitionKey);
for (IdleChannel idleChannel : partition)
close(idleChannel.channel);
}
}
private void close(Channel channel) {
// FIXME pity to have to do this here
Channels.setDiscard(channel);
if (connectionTtlEnabled) {
channelId2Creation.remove(channel.id());
}
Channels.silentlyCloseChannel(channel);
}
flushPartitions会遍历partitions,然后执行predicate.test,为true则执行flushPartition,它将从partitions移除指定的partitionKey,然后遍历idleChannels挨个执行close
getIdleChannelCountPerHost
public Map<String, Long> getIdleChannelCountPerHost() {
return partitions
.values()
.stream()
.flatMap(ConcurrentLinkedDeque::stream)
.map(idle -> idle.getChannel().remoteAddress())
.filter(a -> a.getClass() == InetSocketAddress.class)
.map(a -> (InetSocketAddress) a)
.map(InetSocketAddress::getHostName)
.collect(Collectors.groupingBy(Function.identity(), Collectors.counting()));
}
getIdleChannelCountPerHost则遍历partitions,然后map出remoteAddress获取hostName,然后进行groupBy
PoolLeaseStrategy
public enum PoolLeaseStrategy {
LIFO {
public <E> E lease(Deque<E> d) {
return d.pollFirst();
}
},
FIFO {
public <E> E lease(Deque<E> d) {
return d.pollLast();
}
};
abstract <E> E lease(Deque<E> d);
}
PoolLeaseStrategy是个枚举,定义了LIFO及FIFO两个枚举,LIFO则是对Deque执行pollFirst,FIFO则是对Deque执行pollLast
IdleChannelDetector
private final class IdleChannelDetector implements TimerTask {
private boolean isIdleTimeoutExpired(IdleChannel idleChannel, long now) {
return maxIdleTimeEnabled && now - idleChannel.start >= maxIdleTime;
}
private List<IdleChannel> expiredChannels(ConcurrentLinkedDeque<IdleChannel> partition, long now) {
// lazy create
List<IdleChannel> idleTimeoutChannels = null;
for (IdleChannel idleChannel : partition) {
boolean isIdleTimeoutExpired = isIdleTimeoutExpired(idleChannel, now);
boolean isRemotelyClosed = !Channels.isChannelActive(idleChannel.channel);
boolean isTtlExpired = isTtlExpired(idleChannel.channel, now);
if (isIdleTimeoutExpired || isRemotelyClosed || isTtlExpired) {
LOGGER.debug("Adding Candidate expired Channel {} isIdleTimeoutExpired={} isRemotelyClosed={} isTtlExpired={}", idleChannel.channel, isIdleTimeoutExpired, isRemotelyClosed, isTtlExpired);
if (idleTimeoutChannels == null)
idleTimeoutChannels = new ArrayList<>(1);
idleTimeoutChannels.add(idleChannel);
}
}
return idleTimeoutChannels != null ? idleTimeoutChannels : Collections.emptyList();
}
private List<IdleChannel> closeChannels(List<IdleChannel> candidates) {
// lazy create, only if we hit a non-closeable channel
List<IdleChannel> closedChannels = null;
for (int i = 0; i < candidates.size(); i++) {
// We call takeOwnership here to avoid closing a channel that has just been taken out
// of the pool, otherwise we risk closing an active connection.
IdleChannel idleChannel = candidates.get(i);
if (idleChannel.takeOwnership()) {
LOGGER.debug("Closing Idle Channel {}", idleChannel.channel);
close(idleChannel.channel);
if (closedChannels != null) {
closedChannels.add(idleChannel);
}
} else if (closedChannels == null) {
// first non closeable to be skipped, copy all
// previously skipped closeable channels
closedChannels = new ArrayList<>(candidates.size());
for (int j = 0; j < i; j++)
closedChannels.add(candidates.get(j));
}
}
return closedChannels != null ? closedChannels : candidates;
}
public void run(Timeout timeout) {
if (isClosed.get())
return;
if (LOGGER.isDebugEnabled())
for (Object key : partitions.keySet()) {
int size = partitions.get(key).size();
if (size > 0) {
LOGGER.debug("Entry count for : {} : {}", key, size);
}
}
long start = unpreciseMillisTime();
int closedCount = 0;
int totalCount = 0;
for (ConcurrentLinkedDeque<IdleChannel> partition : partitions.values()) {
// store in intermediate unsynchronized lists to minimize
// the impact on the ConcurrentLinkedDeque
if (LOGGER.isDebugEnabled())
totalCount += partition.size();
List<IdleChannel> closedChannels = closeChannels(expiredChannels(partition, start));
if (!closedChannels.isEmpty()) {
if (connectionTtlEnabled) {
for (IdleChannel closedChannel : closedChannels)
channelId2Creation.remove(closedChannel.channel.id());
}
partition.removeAll(closedChannels);
closedCount += closedChannels.size();
}
}
if (LOGGER.isDebugEnabled()) {
long duration = unpreciseMillisTime() - start;
if (closedCount > 0) {
LOGGER.debug("Closed {} connections out of {} in {} ms", closedCount, totalCount, duration);
}
}
scheduleNewIdleChannelDetector(timeout.task());
}
}
IdleChannelDetector实现了netty的TimerTask接口,其run方法主要是遍历partitions,通过expiredChannels取出过期的IdleChannel,这里isIdleTimeoutExpired、isRemotelyClosed、isTtlExpired都算在内,然后挨个执行takeOwnership及close,再从channelId2Creation及partition中移除,最后再次调度一下IdleChannelDetector
小结
AsyncHttpClient的ChannelPool定义了offer、poll、removeAll、isOpen、destroy、flushPartitions、getIdleChannelCountPerHost方法,它有两个实现类,分别是NoopChannelPool及DefaultChannelPool;DefaultChannelPool基于ConcurrentHashMap实现了ChannelPool接口,主要的参数为connectionTtl、maxIdleTime、cleanerPeriod、poolLeaseStrategy;cleanerPeriod会取connectionTtl、maxIdleTime、传入的cleanerPeriod的最小值;开启connectionTtl或者maxIdleTime的话,会往nettyTimer添加IdleChannelDetector,延后cleanerPeriod时间执行。
poll方法会判断是active,不是的话继续循环lease,而IdleChannelDetector则会定期检查,isIdleTimeoutExpired、isRemotelyClosed、isTtlExpired都会被close,offer的时候还会判断isTtlExpired,这样子来保证连接的活性。
**粗体** _斜体_ [链接](http://example.com) `代码` - 列表 > 引用
。你还可以使用@
来通知其他用户。