从本文开始,咱们来剖析rocketMq音讯接管、散发以及投递流程。
RocketMq音讯解决整个流程如下:
- 音讯接管:音讯接管是指接管producer的音讯,解决类是SendMessageProcessor,将音讯写入到commigLog文件后,接管流程处理完毕;
- 音讯散发:broker解决音讯散发的类是ReputMessageService,它会启动一个线程,一直地将commitLong分到到对应的consumerQueue,这一步操作会写两个文件:consumerQueue与indexFile,写入后,音讯散发流程解决 结束;
- 音讯投递:音讯投递是指将音讯发往consumer的流程,consumer会发动获取音讯的申请,broker收到申请后,调用PullMessageProcessor类解决,从consumerQueue文件获取音讯,返回给consumer后,投递流程处理完毕。
以上就是rocketMq解决音讯的流程了,接下来咱们就从源码来看相干流程的实现。
1. remotingServer的启动流程
在正式剖析接管与投递流程前,咱们来理解下remotingServer的启动。
remotingServer是一个netty服务,他开启了一个端口用来解决producer与consumer的网络申请。
remotingServer是在BrokerController#start中启动的,代码如下:
public void start() throws Exception { // 启动各组件 ... if (this.remotingServer != null) { this.remotingServer.start(); } ... }
持续查看remotingServer的启动流程,进入NettyRemotingServer#start办法:
public void start() { ... ServerBootstrap childHandler = this.serverBootstrap.group(this.eventLoopGroupBoss, this.eventLoopGroupSelector) ... .childHandler(new ChannelInitializer<SocketChannel>() { @Override public void initChannel(SocketChannel ch) throws Exception { ch.pipeline() .addLast(defaultEventExecutorGroup, HANDSHAKE_HANDLER_NAME, handshakeHandler) .addLast(defaultEventExecutorGroup, encoder, new NettyDecoder(), new IdleStateHandler(0, 0, nettyServerConfig.getServerChannelMaxIdleTimeSeconds()), connectionManageHandler, // 解决业务申请的handler serverHandler ); } }); ...}
这就是一个规范的netty服务启动流程了,套路与nameServer的启动是一样的。对于netty的相干内容,这里咱们仅关注pipeline上的channelHandler,在netty中,解决读写申请的操作为一个个ChannelHandler,remotingServer中解决读写申请的ChanelHandler为NettyServerHandler,代码如下:
@ChannelHandler.Sharableclass NettyServerHandler extends SimpleChannelInboundHandler<RemotingCommand> { @Override protected void channelRead0(ChannelHandlerContext ctx, RemotingCommand msg) throws Exception { processMessageReceived(ctx, msg); }}
这块的操作与nameServer对外提供的服务极类似(就是同一个类),最终调用的是NettyRemotingAbstract#processRequestCommand办法:
public void processRequestCommand(final ChannelHandlerContext ctx, final RemotingCommand cmd) { // 依据 code 从 processorTable 获取 Pair final Pair<NettyRequestProcessor, ExecutorService> matched = this.processorTable.get(cmd.getCode()); // 找不到默认值 final Pair<NettyRequestProcessor, ExecutorService> pair = null == matched ? this.defaultRequestProcessor : matched; ... // 从 pair 中拿到 Processor 进行解决 NettyRequestProcessor processor = pair.getObject1(); // 解决申请 RemotingCommand response = processor.processRequest(ctx, cmd); .... }
如果进入源码去看,会发现这个办法十分长,这里省略了异步解决、异样解决及返回值结构等,仅列出了关键步骤:
- 依据code从processorTable拿到对应的Pair
- 从Pair里获取Processor
最终解决申请的就是Processor了。
2. Processor的注册
从下面的剖析中可知, Processor是解决音讯的要害,它是从processorTable中获取的,这个processorTable是啥呢?
processorTable是NettyRemotingAbstract成员变量,外面的内容是BrokerController在初始化时(执行BrokerController#initialize办法)注册的。之前在剖析BrokerController的初始化流程时,就提到过Processor的提供操作,这里再回顾下:
BrokerController的初始化办法initialize会调用 BrokerController#registerProcessor,Processor的注册操作就在这个办法里:
public class BrokerController { private final PullMessageProcessor pullMessageProcessor; /** * 构造方法 */ public BrokerController(...) { // 解决 consumer 拉音讯申请的 this.pullMessageProcessor = new PullMessageProcessor(this); } /** * 注册操作 */ public void registerProcessor() { // SendMessageProcessor SendMessageProcessor sendProcessor = new SendMessageProcessor(this); sendProcessor.registerSendMessageHook(sendMessageHookList); sendProcessor.registerConsumeMessageHook(consumeMessageHookList); // 解决 Processor this.remotingServer.registerProcessor(RequestCode.SEND_MESSAGE, sendProcessor, this.sendMessageExecutor); this.remotingServer.registerProcessor(RequestCode.SEND_MESSAGE_V2, sendProcessor, this.sendMessageExecutor); this.remotingServer.registerProcessor(RequestCode.SEND_BATCH_MESSAGE, sendProcessor, this.sendMessageExecutor); // PullMessageProcessor this.remotingServer.registerProcessor(RequestCode.PULL_MESSAGE, this.pullMessageProcessor, this.pullMessageExecutor); // 省略其余许许多多的Processor注册 ... } ...
须要指明的是,sendProcessor用来解决producer申请过去的音讯,pullMessageProcessor用来解决consumer拉取音讯的申请。
3. 接管producer音讯
理解完remotingServer的启动与Processor的注册内容后,接下来咱们就能够剖析接管producer音讯的流程了。
producer发送音讯到broker时,发送的申请code为SEND_MESSAGE(RocketMQ源码5-producer 同步发送和单向发送 第1.4大节),依据下面的剖析,当音讯过去时,会应用NettyServerHandler这个ChannelHandler来解决,之后会调用到NettyRemotingAbstract#processRequestCommand办法。
在NettyRemotingAbstract#processRequestCommand办法中,会依据音讯的code获取对应的Processor来解决,从Processor的注册流程来看,解决该SEND_MESSAGE的Processor为SendMessageProcessor,咱们进入SendMessageProcessor#processRequest看看它的流程:
public RemotingCommand processRequest(ChannelHandlerContext ctx, RemotingCommand request) throws RemotingCommandException { RemotingCommand response = null; try { // broker解决接管音讯 response = asyncProcessRequest(ctx, request).get(); } catch (InterruptedException | ExecutionException e) { log.error("process SendMessage error, request : " + request.toString(), e); } return response;}
没干啥事,一路跟上来,间接看一般音讯的流程,进入SendMessageProcessor#asyncSendMessage办法:
private CompletableFuture<RemotingCommand> asyncSendMessage(ChannelHandlerContext ctx, RemotingCommand request, SendMessageContext mqtraceContext, SendMessageRequestHeader requestHeader) { final RemotingCommand response = preSend(ctx, request, requestHeader); final SendMessageResponseHeader responseHeader = (SendMessageResponseHeader)response.readCustomHeader(); if (response.getCode() != -1) { return CompletableFuture.completedFuture(response); } final byte[] body = request.getBody(); int queueIdInt = requestHeader.getQueueId(); TopicConfig topicConfig = this.brokerController.getTopicConfigManager() .selectTopicConfig(requestHeader.getTopic()); // 如果没指定队列,就随机指定一个队列 if (queueIdInt < 0) { queueIdInt = randomQueueId(topicConfig.getWriteQueueNums()); } // 将音讯包装为 MessageExtBrokerInner MessageExtBrokerInner msgInner = new MessageExtBrokerInner(); msgInner.setTopic(requestHeader.getTopic()); msgInner.setQueueId(queueIdInt); // 省略解决 msgInner 的流程 ... CompletableFuture<PutMessageResult> putMessageResult = null; Map<String, String> origProps = MessageDecoder .string2messageProperties(requestHeader.getProperties()); String transFlag = origProps.get(MessageConst.PROPERTY_TRANSACTION_PREPARED); // 发送事务音讯 if (transFlag != null && Boolean.parseBoolean(transFlag)) { ... // 发送事务音讯 putMessageResult = this.brokerController.getTransactionalMessageService() .asyncPrepareMessage(msgInner); } else { // 发送一般音讯 putMessageResult = this.brokerController.getMessageStore().asyncPutMessage(msgInner); } return handlePutMessageResultFuture(putMessageResult, response, request, msgInner, responseHeader, mqtraceContext, ctx, queueIdInt);}
这个办法是在筹备音讯的发送数据,所做的工作如下:
- 如果没指定队列,就随机指定一个队列,个别状况下不会给音讯指定队列的,但如果要发送程序音讯,就须要指定队列了,这点前面再剖析。
- 结构MessageExtBrokerInner对象,就是将producer上送的音讯包装下,加上一些额定的信息,如音讯标识msgId、发送工夫、topic、queue等。
- 发送音讯,这里只是分为两类:事务音讯与一般音讯,这里咱们次要关注一般音讯,事务音讯前面再剖析。
进入一般音讯的发送办法DefaultMessageStore#asyncPutMessage:
public CompletableFuture<PutMessageResult> asyncPutMessage(MessageExtBrokerInner msg) { ... // 保留到 commitLog CompletableFuture<PutMessageResult> putResultFuture = this.commitLog.asyncPutMessage(msg); ...}
3.1 commitLog写入原理
一个broker逻辑上对应着一个commitLog,你能够把它看作一个大文件,而后这个broker收到的所有音讯都写到这个外面,然而物理上ROCKET_HOME/commitlog/00000000000000000000这个门路存储的,它是由若干个文件组成的,每个文件默认大小是1G,而后每个文件都对应这个一个MappedFile,00000000000000000000 就是第一个MappedFile对应的物理文件,每个文件的文件名就是在commitLog外面的一个其实offset,第二个文件名就是00000000001073741824,也就是上一个MappedFile文件起始offset加上每个文件的大小,这个MappedFile就是RocketMQ的黑科技,应用了内存映射技术来进步文件的访问速度与写入速度,而后都是采纳追加写的形式进步写入速度。
咱们间接看官网的形容(链接:github.com/apache/rock…):
rocketMq 音讯存储架构图
音讯存储架构图中次要有上面三个跟音讯存储相干的文件形成。
(1) CommitLog:音讯主体以及元数据的存储主体,存储Producer端写入的音讯主体内容,音讯内容不是定长的。单个文件大小默认1G ,文件名长度为20位,右边补零,残余为起始偏移量,比方00000000000000000000代表了第一个文件,起始偏移量为0,文件大小为1G=1073741824;当第一个文件写满了,第二个文件为00000000001073741824,起始偏移量为1073741824,以此类推。音讯次要是程序写入日志文件,当文件满了,写入下一个文件;
(2) ConsumeQueue:音讯生产队列,引入的目标次要是进步音讯生产的性能,因为RocketMQ是基于主题topic的订阅模式,音讯生产是针对主题进行的,如果要遍历commitlog文件中依据topic检索音讯是十分低效的。Consumer即可依据ConsumeQueue来查找待生产的音讯。其中,ConsumeQueue(逻辑生产队列)作为生产音讯的索引,保留了指定Topic下的队列音讯在CommitLog中的起始物理偏移量offset,音讯大小size和音讯Tag的HashCode值。consumequeue文件能够看成是基于topic的commitlog索引文件,故consumequeue文件夹的组织形式如下:topic/queue/file三层组织构造,具体存储门路为:$HOME/store/consumequeue/{topic}/{queueId}/{fileName}。同样consumequeue文件采取定长设计,每一个条目共20个字节,别离为8字节的commitlog物理偏移量、4字节的音讯长度、8字节tag hashcode,单个文件由30W个条目组成,能够像数组一样随机拜访每一个条目,每个ConsumeQueue文件大小约5.72M;
(3) IndexFile:IndexFile(索引文件)提供了一种能够通过key或工夫区间来查问音讯的办法。Index文件的存储地位是:HOME\store\index{fileName},文件名fileName是以创立时的工夫戳命名的,固定的单个IndexFile文件大小约为400M,一个IndexFile能够保留 2000W个索引,IndexFile的底层存储设计为在文件系统中实现HashMap构造,故rocketmq的索引文件其底层实现为hash索引。
在下面的RocketMQ的音讯存储整体架构图中能够看出,RocketMQ采纳的是混合型的存储构造,即为Broker单个实例下所有的队列共用一个日志数据文件(即为CommitLog)来存储。RocketMQ的混合型存储构造(多个Topic的音讯实体内容都存储于一个CommitLog中)针对Producer和Consumer别离采纳了数据和索引局部相拆散的存储构造,Producer发送音讯至Broker端,而后Broker端应用同步或者异步的形式对音讯刷盘长久化,保留至CommitLog中。只有音讯被刷盘长久化至磁盘文件CommitLog中,那么Producer发送的音讯就不会失落。
正因为如此,Consumer也就必定有机会去生产这条音讯。当无奈拉取到音讯后,能够等下一次音讯拉取,同时服务端也反对长轮询模式,如果一个音讯拉取申请未拉取到音讯,Broker容许期待30s的工夫,只有这段时间内有新音讯达到,将间接返回给生产端。这里,RocketMQ的具体做法是,应用Broker端的后盾服务线程—ReputMessageService不停地散发申请并异步构建ConsumeQueue(逻辑生产队列)和IndexFile(索引文件)数据。
3.2 CommitLog#asyncPutMessage
持续进入CommitLog#asyncPutMessage办法,这个办法有点长,咱们分局部:
public CompletableFuture<PutMessageResult> asyncPutMessage(final MessageExtBrokerInner msg) { // Set the storage time 设置存储工夫 msg.setStoreTimestamp(System.currentTimeMillis()); // Set the message body BODY CRC (consider the most appropriate setting // on the client) // 设置crc msg.setBodyCRC(UtilAll.crc32(msg.getBody())); // Back to Results AppendMessageResult result = null; StoreStatsService storeStatsService = this.defaultMessageStore.getStoreStatsService(); String topic = msg.getTopic(); int queueId = msg.getQueueId(); // 获取事务状态 final int tranType = MessageSysFlag.getTransactionValue(msg.getSysFlag()); // 事务 if (tranType == MessageSysFlag.TRANSACTION_NOT_TYPE || tranType == MessageSysFlag.TRANSACTION_COMMIT_TYPE) { // Delay Delivery 延时音讯的解决 if (msg.getDelayTimeLevel() > 0) { if (msg.getDelayTimeLevel() > this.defaultMessageStore.getScheduleMessageService().getMaxDelayLevel()) { msg.setDelayTimeLevel(this.defaultMessageStore.getScheduleMessageService().getMaxDelayLevel()); } // 设置提早队列 topic = TopicValidator.RMQ_SYS_SCHEDULE_TOPIC; queueId = ScheduleMessageService.delayLevel2QueueId(msg.getDelayTimeLevel()); // Backup real topic, queueId MessageAccessor.putProperty(msg, MessageConst.PROPERTY_REAL_TOPIC, msg.getTopic()); MessageAccessor.putProperty(msg, MessageConst.PROPERTY_REAL_QUEUE_ID, String.valueOf(msg.getQueueId())); msg.setPropertiesString(MessageDecoder.messageProperties2String(msg.getProperties())); msg.setTopic(topic); msg.setQueueId(queueId); } } ...
这一部分其实就是从msg中获取一些信息,判断解决一下这个延时音讯。
long elapsedTimeInLock = 0;MappedFile unlockMappedFile = null;// 获取最初一个 MappedFileMappedFile mappedFile = this.mappedFileQueue.getLastMappedFile();// 获取写入锁putMessageLock.lock(); //spin or ReentrantLock ,depending on store configtry { long beginLockTimestamp = this.defaultMessageStore.getSystemClock().now(); // 开始在锁里的工夫 this.beginTimeInLock = beginLockTimestamp; // Here settings are stored timestamp, in order to ensure an orderly // global // 设置写入的工夫戳,确保它是有序的, msg.setStoreTimestamp(beginLockTimestamp); // 判断MappedFile 是否是 null 或者是否是满了 if (null == mappedFile || mappedFile.isFull()) { mappedFile = this.mappedFileQueue.getLastMappedFile(0); // Mark: NewFile may be cause noise } if (null == mappedFile) { log.error("create mapped file1 error, topic: " + msg.getTopic() + " clientAddr: " + msg.getBornHostString()); beginTimeInLock = 0; return CompletableFuture.completedFuture(new PutMessageResult(PutMessageStatus.CREATE_MAPEDFILE_FAILED, null)); } // todo 往mappedFile追加音讯 result = mappedFile.appendMessage(msg, this.appendMessageCallback); ...
这一部分就比拟重要了,首先是从mappedFileQueue中获取最初一个MappedFile,这个就是拿汇合最初一个元素,因为都是有序的,最初一个元素就是最初一个MappedFile,接着就是获取锁了,这个锁也是比拟有考究的,能够设置应用ReentrantLock 也能够设置应用cas,默认是应用cas,接着就是设置beginTimeInLock这个变量了,这个变量咱们在判断os page cache忙碌的时候说过,就是获取到锁的一个工夫戳,在开释锁之前会重置成0,接着就是判断mappedFile是不是null或者是不是满了,如果是的话就要新建一个了。
接着就是最最最重要的了 往mappedFile中追加音讯,
mappedFile.appendMessage
/** * 将音讯追加到MappedFile文件中 */public AppendMessageResult appendMessagesInner(final MessageExt messageExt, final AppendMessageCallback cb) { assert messageExt != null; assert cb != null; // 获取MappedFile以后文件写指针 int currentPos = this.wrotePosition.get(); // 如果currentPos小于文件大小 if (currentPos < this.fileSize) { ByteBuffer byteBuffer = writeBuffer != null ? writeBuffer.slice() : this.mappedByteBuffer.slice(); byteBuffer.position(currentPos); AppendMessageResult result; // 单个音讯 if (messageExt instanceof MessageExtBrokerInner) { // todo result = cb.doAppend(this.getFileFromOffset(), byteBuffer, this.fileSize - currentPos, (MessageExtBrokerInner) messageExt); // 批量音讯 } else if (messageExt instanceof MessageExtBatch) { result = cb.doAppend(this.getFileFromOffset(), byteBuffer, this.fileSize - currentPos, (MessageExtBatch) messageExt); } else { return new AppendMessageResult(AppendMessageStatus.UNKNOWN_ERROR); } this.wrotePosition.addAndGet(result.getWroteBytes()); this.storeTimestamp = result.getStoreTimestamp(); return result; } // 如果currentPos大于或等于文件大小,表明文件已写满,抛出异样 log.error("MappedFile.appendMessage return null, wrotePosition: {} fileSize: {}", currentPos, this.fileSize); return new AppendMessageResult(AppendMessageStatus.UNKNOWN_ERROR);}
这里首先获取了一下这个mappedFile写到哪个地位了,它这个地位是从0开始的,而后判断一下以后地位与文件大小做比照,要是大于的就超了文件大小了,接着是获取writerbuffer因为这里是没有开启transientStorePool的,所以它是个空的,就会应用mmapedByteBuffer,接着就是调用回调的doAppend追加音讯了,咱们看下它的参数, 第一个是开始offset,这个offset是commitlog的一个offset,举个例子,第一个MappedFile的开始offset是0,而后一个MappedFile 的大小是1g,而后第二个MappedFile就得从1073741824(1g)开始了,第二个参数是bytebuffer,这个不必多说,第三个是这个MappedFile还空多少字节没用,第四个就是音讯了。
咱们来看下这个doAppend办法,这个也有点长,咱们须要离开看下:
/** * // 只是将音讯追加到内存中 * @param fileFromOffset 文件的第一个偏移量(就是MappedFile是从哪个中央开始的) */public AppendMessageResult doAppend(final long fileFromOffset, final ByteBuffer byteBuffer, final int maxBlank, final MessageExtBrokerInner msgInner) { // STORETIMESTAMP + STOREHOSTADDRESS + OFFSET <br> // PHY OFFSET long wroteOffset = fileFromOffset + byteBuffer.position(); int sysflag = msgInner.getSysFlag(); int bornHostLength = (sysflag & MessageSysFlag.BORNHOST_V6_FLAG) == 0 ? 4 + 4 : 16 + 4; int storeHostLength = (sysflag & MessageSysFlag.STOREHOSTADDRESS_V6_FLAG) == 0 ? 4 + 4 : 16 + 4; ByteBuffer bornHostHolder = ByteBuffer.allocate(bornHostLength); ByteBuffer storeHostHolder = ByteBuffer.allocate(storeHostLength); this.resetByteBuffer(storeHostHolder, storeHostLength); // 创立全局惟一音讯id String msgId; if ((sysflag & MessageSysFlag.STOREHOSTADDRESS_V6_FLAG) == 0) { msgId = MessageDecoder.createMessageId(this.msgIdMemory, msgInner.getStoreHostBytes(storeHostHolder), wroteOffset); } else { msgId = MessageDecoder.createMessageId(this.msgIdV6Memory, msgInner.getStoreHostBytes(storeHostHolder), wroteOffset); } // Record ConsumeQueue information keyBuilder.setLength(0); keyBuilder.append(msgInner.getTopic()); keyBuilder.append('-'); keyBuilder.append(msgInner.getQueueId()); String key = keyBuilder.toString(); // 获取该音讯在音讯队列的物理偏移量 Long queueOffset = CommitLog.this.topicQueueTable.get(key); if (null == queueOffset) { queueOffset = 0L; CommitLog.this.topicQueueTable.put(key, queueOffset); } // Transaction messages that require special handling final int tranType = MessageSysFlag.getTransactionValue(msgInner.getSysFlag()); switch (tranType) { // Prepared and Rollback message is not consumed, will not enter the // consumer queue case MessageSysFlag.TRANSACTION_PREPARED_TYPE: case MessageSysFlag.TRANSACTION_ROLLBACK_TYPE: queueOffset = 0L; break; case MessageSysFlag.TRANSACTION_NOT_TYPE: case MessageSysFlag.TRANSACTION_COMMIT_TYPE: default: break; }
这一部分次要就是 计算了一下这个音讯写在commitlog中的一个offset,接着就是生成一个msgId,而后依据topic 与queueId从缓存中获取了一下这个queueId对应的一个queue的offset,这个其实就是增加一个音讯加1,而后就是事务的货色了,如果有事务,而后还在筹备阶段或者回滚阶段,就将queue offset 设置成0,再往下其实就是解决音讯,而后写到buffer中了。
/** * Serialize message */final byte[] propertiesData = msgInner.getPropertiesString() == null ? null : msgInner.getPropertiesString().getBytes(MessageDecoder.CHARSET_UTF8);final int propertiesLength = propertiesData == null ? 0 : propertiesData.length;if (propertiesLength > Short.MAX_VALUE) { log.warn("putMessage message properties length too long. length={}", propertiesData.length); return new AppendMessageResult(AppendMessageStatus.PROPERTIES_SIZE_EXCEEDED);}final byte[] topicData = msgInner.getTopic().getBytes(MessageDecoder.CHARSET_UTF8);final int topicLength = topicData.length;final int bodyLength = msgInner.getBody() == null ? 0 : msgInner.getBody().length;// todo 计算音讯总长度final int msgLen = calMsgLength(msgInner.getSysFlag(), bodyLength, topicLength, propertiesLength);// Exceeds the maximum message if (msgLen > this.maxMessageSize) { // 最大音讯长度 65536 CommitLog.log.warn("message size exceeded, msg total size: " + msgLen + ", msg body size: " + bodyLength + ", maxMessageSize: " + this.maxMessageSize); return new AppendMessageResult(AppendMessageStatus.MESSAGE_SIZE_EXCEEDED);}...
这里首先是获取了一下音讯外面的properties,将它转成字节数组,计算了一下长度,接着就是将topic转成字节数据,计算了一下长度,获取了一下body的长度,就是你往Message塞得内容长度,重点来,计算 音讯的总长度,而后判断一下长度是否超长。其中calMsgLength如下:
// 计算音讯长度 protected static int calMsgLength(int sysFlag, int bodyLength, int topicLength, int propertiesLength) { int bornhostLength = (sysFlag & MessageSysFlag.BORNHOST_V6_FLAG) == 0 ? 8 : 20; int storehostAddressLength = (sysFlag & MessageSysFlag.STOREHOSTADDRESS_V6_FLAG) == 0 ? 8 : 20; final int msgLen = 4 //TOTALSIZE 音讯条目总长度,4字节 + 4 //MAGICCODE 魔数,4字节。固定值0xdaa320a7 + 4 //BODYCRC 音讯体的crc校验码,4字节 + 4 //QUEUEID 音讯生产队列ID,4字节 + 4 //FLAG 音讯标记,RocketMQ对其不做解决,供应用程序应用, 默认4字节 + 8 //QUEUEOFFSET 音讯在ConsumeQuene文件中的物理偏移量,8字节。 + 8 //PHYSICALOFFSET 音讯在CommitLog文件中的物理偏移量,8字节 + 4 //SYSFLAG 音讯零碎标记,例如是否压缩、是否是事务音讯 等,4字节 + 8 //BORNTIMESTAMP 音讯生产者调用音讯发送API的工夫戳,8字 节 + bornhostLength //BORNHOST 音讯发送者IP、端口号,8字节 + 8 //STORETIMESTAMP 音讯存储工夫戳,8字节 + storehostAddressLength //STOREHOSTADDRESS Broker服务器IP+端口号,8字节 + 4 //RECONSUMETIMES 音讯重试次数,4字节 + 8 //Prepared Transaction Offset 事务音讯的物理偏移量,8 字节。 + 4 // 音讯体长度,4字节 + (bodyLength > 0 ? bodyLength : 0) // BODY 音讯体内容,长度为bodyLenth中存储的值 + 1 // 主题存储长度,1字节,示意主题名称不能超过255个字符。 + topicLength //TOPIC 主题,长度为TopicLength中存储的值 + 2 // 音讯属性长度,2字节,示意音讯属性长 度不能超过65536个字符。 + (propertiesLength > 0 ? propertiesLength : 0) //propertiesLength 音讯属性,长度为PropertiesLength中存储的 值 + 0; return msgLen; }
持续:
...// todo 音讯长度+END_FILE_MIN_BLANK_LENGTH 大于commitLog的闲暇空间,则返回END_OF_FILEif ((msgLen + END_FILE_MIN_BLANK_LENGTH) > maxBlank) { this.resetByteBuffer(this.msgStoreItemMemory, maxBlank); // 1 TOTALSIZE 4字节存储以后文件的残余空间 this.msgStoreItemMemory.putInt(maxBlank); // 2 MAGICCODE 4字节存储魔数 this.msgStoreItemMemory.putInt(CommitLog.BLANK_MAGIC_CODE); // 3 The remaining space may be any value // Here the length of the specially set maxBlank final long beginTimeMills = CommitLog.this.defaultMessageStore.now(); byteBuffer.put(this.msgStoreItemMemory.array(), 0, maxBlank); return new AppendMessageResult(AppendMessageStatus.END_OF_FILE, wroteOffset, maxBlank, msgId, msgInner.getStoreTimestamp(), queueOffset, CommitLog.this.defaultMessageStore.now() - beginTimeMills);}...
判断剩下的空间能不能放开,如果放不开的话,就塞上一个完结的货色,8个字节是正经的,剩下的随便,而后返回文件满了的状态。
...// Initialization of storage space// 初始化存储空间this.resetByteBuffer(msgStoreItemMemory, msgLen);// 1 TOTALSIZEthis.msgStoreItemMemory.putInt(msgLen);// 2 MAGICCODEthis.msgStoreItemMemory.putInt(CommitLog.MESSAGE_MAGIC_CODE);// 3 BODYCRCthis.msgStoreItemMemory.putInt(msgInner.getBodyCRC());// 4 QUEUEIDthis.msgStoreItemMemory.putInt(msgInner.getQueueId());// 5 FLAGthis.msgStoreItemMemory.putInt(msgInner.getFlag());// 6 QUEUEOFFSETthis.msgStoreItemMemory.putLong(queueOffset);// 7 PHYSICALOFFSETthis.msgStoreItemMemory.putLong(fileFromOffset + byteBuffer.position());// 8 SYSFLAGthis.msgStoreItemMemory.putInt(msgInner.getSysFlag());// 9 BORNTIMESTAMPthis.msgStoreItemMemory.putLong(msgInner.getBornTimestamp());// 10 BORNHOSTthis.resetByteBuffer(bornHostHolder, bornHostLength);this.msgStoreItemMemory.put(msgInner.getBornHostBytes(bornHostHolder));// 11 STORETIMESTAMPthis.msgStoreItemMemory.putLong(msgInner.getStoreTimestamp());// 12 STOREHOSTADDRESSthis.resetByteBuffer(storeHostHolder, storeHostLength);this.msgStoreItemMemory.put(msgInner.getStoreHostBytes(storeHostHolder));// 13 RECONSUMETIMESthis.msgStoreItemMemory.putInt(msgInner.getReconsumeTimes());// 14 Prepared Transaction Offsetthis.msgStoreItemMemory.putLong(msgInner.getPreparedTransactionOffset());// 15 BODYthis.msgStoreItemMemory.putInt(bodyLength);if (bodyLength > 0) this.msgStoreItemMemory.put(msgInner.getBody());// 16 TOPICthis.msgStoreItemMemory.put((byte) topicLength);this.msgStoreItemMemory.put(topicData);// 17 PROPERTIESthis.msgStoreItemMemory.putShort((short) propertiesLength);if (propertiesLength > 0) this.msgStoreItemMemory.put(propertiesData);final long beginTimeMills = CommitLog.this.defaultMessageStore.now();// Write messages to the queue bufferbyteBuffer.put(this.msgStoreItemMemory.array(), 0, msgLen);...
这个就是封装音讯了,最初将音讯放到byteBuffer中。
...// 创立AppendMessageResult AppendMessageResult result = new AppendMessageResult(AppendMessageStatus.PUT_OK, wroteOffset, msgLen, msgId, msgInner.getStoreTimestamp(), queueOffset, CommitLog.this.defaultMessageStore.now() - beginTimeMills); switch (tranType) { case MessageSysFlag.TRANSACTION_PREPARED_TYPE: case MessageSysFlag.TRANSACTION_ROLLBACK_TYPE: break; case MessageSysFlag.TRANSACTION_NOT_TYPE: case MessageSysFlag.TRANSACTION_COMMIT_TYPE: // The next update ConsumeQueue information // 更新音讯队列的逻辑偏移量 CommitLog.this.topicQueueTable.put(key, ++queueOffset); break; default: break; } return result;}
最初就是封装追加音讯的后果是put_ok,而后更新queue offset ,其实就是+1。
接下来咱们回过头来看下appendMessagesInner的后半局部,
...this.wrotePosition.addAndGet(result.getWroteBytes());this.storeTimestamp = result.getStoreTimestamp();return result;
这里其实就是更新了一下 这个MappedFile 写到哪个中央了,更新了下写入工夫。
回到commitLog的putMessage办法:
... // todo 往mappedFile追加音讯 result = mappedFile.appendMessage(msg, this.appendMessageCallback); switch (result.getStatus()) { case PUT_OK: break; case END_OF_FILE: unlockMappedFile = mappedFile; // Create a new file, re-write the message mappedFile = this.mappedFileQueue.getLastMappedFile(0); if (null == mappedFile) { // XXX: warn and notify me log.error("create mapped file2 error, topic: " + msg.getTopic() + " clientAddr: " + msg.getBornHostString()); beginTimeInLock = 0; return CompletableFuture.completedFuture(new PutMessageResult(PutMessageStatus.CREATE_MAPEDFILE_FAILED, result)); } result = mappedFile.appendMessage(msg, this.appendMessageCallback); break; case MESSAGE_SIZE_EXCEEDED: case PROPERTIES_SIZE_EXCEEDED: beginTimeInLock = 0; return CompletableFuture.completedFuture(new PutMessageResult(PutMessageStatus.MESSAGE_ILLEGAL, result)); case UNKNOWN_ERROR: beginTimeInLock = 0; return CompletableFuture.completedFuture(new PutMessageResult(PutMessageStatus.UNKNOWN_ERROR, result)); default: beginTimeInLock = 0; return CompletableFuture.completedFuture(new PutMessageResult(PutMessageStatus.UNKNOWN_ERROR, result)); } elapsedTimeInLock = this.defaultMessageStore.getSystemClock().now() - beginLockTimestamp; beginTimeInLock = 0;} finally { putMessageLock.unlock();}...
这里追加实现了,就须要判断追加状态了,如果是那种MappedFile放不开音讯的状况,它会从新获取一个MappedFile,而后从新追加,在开释锁之前,它还会将beginTimeInLock这个字段重置为0;
...if (elapsedTimeInLock > 500) { log.warn("[NOTIFYME]putMessage in lock cost time(ms)={}, bodyLength={} AppendMessageResult={}", elapsedTimeInLock, msg.getBody().length, result); } if (null != unlockMappedFile && this.defaultMessageStore.getMessageStoreConfig().isWarmMapedFileEnable()) { this.defaultMessageStore.unlockMappedFile(unlockMappedFile); } PutMessageResult putMessageResult = new PutMessageResult(PutMessageStatus.PUT_OK, result); // Statistics storeStatsService.getSinglePutMessageTopicTimesTotal(msg.getTopic()).incrementAndGet(); storeStatsService.getSinglePutMessageTopicSizeTotal(topic).addAndGet(result.getWroteBytes()); // todo 音讯首先进入pagecache,而后执行刷盘操作, CompletableFuture<PutMessageStatus> flushResultFuture = submitFlushRequest(result, msg); // todo 接着调用submitReplicaRequest办法将音讯提交到HaService,进行数据复制 CompletableFuture<PutMessageStatus> replicaResultFuture = submitReplicaRequest(result, msg); // todo 这里应用了ComplateFuture的thenCombine办法,将刷盘、复制当成一 // todo 个联结工作执行,这里设置音讯追加的最终状态 return flushResultFuture.thenCombine(replicaResultFuture, (flushStatus, replicaStatus) -> { if (flushStatus != PutMessageStatus.PUT_OK) { putMessageResult.setPutMessageStatus(flushStatus); } if (replicaStatus != PutMessageStatus.PUT_OK) { putMessageResult.setPutMessageStatus(replicaStatus); if (replicaStatus == PutMessageStatus.FLUSH_SLAVE_TIMEOUT) { log.error("do sync transfer other node, wait return, but failed, topic: {} tags: {} client address: {}", msg.getTopic(), msg.getTags(), msg.getBornHostNameString()); } } return putMessageResult; });}
判断了一下耗时,如果是大于500ms的话,打印正告,封装put音讯的后果,统计store,能够看到前面调用了2个办法,一个是刷盘的,一个是同步音讯的,咱们这里要看下这个刷盘动作:
public CompletableFuture<PutMessageStatus> submitFlushRequest(AppendMessageResult result, MessageExt messageExt) { // Synchronization flush 同步刷盘 if (FlushDiskType.SYNC_FLUSH == this.defaultMessageStore.getMessageStoreConfig().getFlushDiskType()) { final GroupCommitService service = (GroupCommitService) this.flushCommitLogService; if (messageExt.isWaitStoreMsgOK()) { // 构建GroupCommitRequest同步工作并提交到GroupCommitRequest GroupCommitRequest request = new GroupCommitRequest(result.getWroteOffset() + result.getWroteBytes(), this.defaultMessageStore.getMessageStoreConfig().getSyncFlushTimeout()); // 刷盘申请 service.putRequest(request); return request.future(); } else { service.wakeup(); return CompletableFuture.completedFuture(PutMessageStatus.PUT_OK); } } // Asynchronous flush 异步刷盘 这个就是靠os else { if (!this.defaultMessageStore.getMessageStoreConfig().isTransientStorePoolEnable()) { flushCommitLogService.wakeup(); } else { commitLogService.wakeup(); } return CompletableFuture.completedFuture(PutMessageStatus.PUT_OK); }}
如果broker配置的SYNC_FLUSH 并且是个同步音讯,这个时候就会创立一个刷盘申请,而后提交刷盘申请,这个时候会等着刷盘实现,默认就是5s。
接着就是到存储器的putMessage办法的后半局部了:
...// todo 存储音讯CompletableFuture<PutMessageResult> putResultFuture = this.commitLog.asyncPutMessage(msg);putResultFuture.thenAccept((result) -> { long elapsedTime = this.getSystemClock().now() - beginTime; if (elapsedTime > 500) { log.warn("putMessage not in lock elapsed time(ms)={}, bodyLength={}", elapsedTime, msg.getBody().length); } // 记录状态 this.storeStatsService.setPutMessageEntireTimeMax(elapsedTime); if (null == result || !result.isOk()) { // 记录状态 this.storeStatsService.getPutMessageFailedTimes().incrementAndGet(); }});return putResultFuture;
commitlog存入音讯之后,咱们这块也就算是实现了,最初就是回到那个processor,而后将put后果写入对应的channel给返回去,通知音讯生产者音讯写入后果 。音讯存储其实就是找对应的MappedFile,依照肯定的格局往文件外面写入,须要留神的是内存映射文件。
这里附一张音讯存储字段存储程序与字段长度的图:
4. 总结
本文次要剖析了 broker 接管producer音讯的流程,流程如下:
- 解决音讯接管的底层服务为 netty,在BrokerController#start办法中启动
- netty服务中,解决音讯接管的channelHandler为NettyServerHandler,最终会调用SendMessageProcessor#processRequest来解决音讯接管
- 音讯接管流程的最初,MappedFile#appendMessage(...)办法会将音讯内容写入到commitLog文件中。