RocketMQ 源码分析 事务消息
2024-06-09 07:00:54
为什么80%的码农都做不了架构师？>>> 
1. 概述
必须必须必须 前置阅读内容：
- 《事务消息（阿里云）》
2. 事务消息发送
2.1 Producer 发送事务消息
- 活动图如下（结合
核心代码理解）：
- 实现代码如下：
1: // 【DefaultMQProducerImpl.java】2: /**3: * 发送事务消息4: *5: * @param msg 消息6: * @param tranExecuter 【本地事务】执行器7: * @param arg 【本地事务】执行器参数8: * @return 事务发送结果9: * @throws MQClientException 当 Client 发生异常时10: */11: public TransactionSendResult sendMessageInTransaction(final Message msg, final LocalTransactionExecuter tranExecuter, final Object arg)12: throws MQClientException {13: if (null == tranExecuter) {14: throw new MQClientException("tranExecutor is null", null);15: }16: Validators.checkMessage(msg, this.defaultMQProducer);17: 18: // 发送【Half消息】19: SendResult sendResult;20: MessageAccessor.putProperty(msg, MessageConst.PROPERTY_TRANSACTION_PREPARED, "true");21: MessageAccessor.putProperty(msg, MessageConst.PROPERTY_PRODUCER_GROUP, this.defaultMQProducer.getProducerGroup());22: try {23: sendResult = this.send(msg);24: } catch (Exception e) {25: throw new MQClientException("send message Exception", e);26: }27: 28: // 处理发送【Half消息】结果29: LocalTransactionState localTransactionState = LocalTransactionState.UNKNOW;30: Throwable localException = null;31: switch (sendResult.getSendStatus()) {32: // 发送【Half消息】成功，执行【本地事务】逻辑33: case SEND_OK: {34: try {35: if (sendResult.getTransactionId() != null) { // 事务编号，目前开源版本暂时没用到，猜想ONS在使用36: msg.putUserProperty("__transactionId__", sendResult.getTransactionId());37: }38: 39: // 执行【本地事务】逻辑40: localTransactionState = tranExecuter.executeLocalTransactionBranch(msg, arg);41: if (null == localTransactionState) {42: localTransactionState = LocalTransactionState.UNKNOW;43: }44: 45: if (localTransactionState != LocalTransactionState.COMMIT_MESSAGE) {46: log.info("executeLocalTransactionBranch return {}", localTransactionState);47: log.info(msg.toString());48: }49: } catch (Throwable e) {50: log.info("executeLocalTransactionBranch exception", e);51: log.info(msg.toString());52: localException = e;53: }54: }55: break;56: // 发送【Half消息】失败，标记【本地事务】状态为回滚57: case FLUSH_DISK_TIMEOUT:58: case FLUSH_SLAVE_TIMEOUT:59: case SLAVE_NOT_AVAILABLE:60: localTransactionState = LocalTransactionState.ROLLBACK_MESSAGE;61: break;62: default:63: break;64: }65: 66: // 结束事务：提交消息 COMMIT / ROLLBACK67: try {68: this.endTransaction(sendResult, localTransactionState, localException);69: } catch (Exception e) {70: log.warn("local transaction execute " + localTransactionState + ", but end broker transaction failed", e);71: }72: 73: // 返回【事务发送结果】74: TransactionSendResult transactionSendResult = new TransactionSendResult();75: transactionSendResult.setSendStatus(sendResult.getSendStatus());76: transactionSendResult.setMessageQueue(sendResult.getMessageQueue());77: transactionSendResult.setMsgId(sendResult.getMsgId());78: transactionSendResult.setQueueOffset(sendResult.getQueueOffset());79: transactionSendResult.setTransactionId(sendResult.getTransactionId());80: transactionSendResult.setLocalTransactionState(localTransactionState);81: return transactionSendResult;82: }83: 84: /**85: * 结束事务：提交消息 COMMIT / ROLLBACK86: *87: * @param sendResult 发送【Half消息】结果88: * @param localTransactionState 【本地事务】状态89: * @param localException 执行【本地事务】逻辑产生的异常90: * @throws RemotingException 当远程调用发生异常时91: * @throws MQBrokerException 当 Broker 发生异常时92: * @throws InterruptedException 当线程中断时93: * @throws UnknownHostException 当解码消息编号失败是94: */95: public void endTransaction(//96: final SendResult sendResult, //97: final LocalTransactionState localTransactionState, //98: final Throwable localException) throws RemotingException, MQBrokerException, InterruptedException, UnknownHostException {99: // 解码消息编号
100: final MessageId id;
101: if (sendResult.getOffsetMsgId() != null) {
102: id = MessageDecoder.decodeMessageId(sendResult.getOffsetMsgId());
103: } else {
104: id = MessageDecoder.decodeMessageId(sendResult.getMsgId());
105: }
106:
107: // 创建请求
108: String transactionId = sendResult.getTransactionId();
109: final String brokerAddr = this.mQClientFactory.findBrokerAddressInPublish(sendResult.getMessageQueue().getBrokerName());
110: EndTransactionRequestHeader requestHeader = new EndTransactionRequestHeader();
111: requestHeader.setTransactionId(transactionId);
112: requestHeader.setCommitLogOffset(id.getOffset());
113: switch (localTransactionState) {
114: case COMMIT_MESSAGE:
115: requestHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_COMMIT_TYPE);
116: break;
117: case ROLLBACK_MESSAGE:
118: requestHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_ROLLBACK_TYPE);
119: break;
120: case UNKNOW:
121: requestHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_NOT_TYPE);
122: break;
123: default:
124: break;
125: }
126: requestHeader.setProducerGroup(this.defaultMQProducer.getProducerGroup());
127: requestHeader.setTranStateTableOffset(sendResult.getQueueOffset());
128: requestHeader.setMsgId(sendResult.getMsgId());
129: String remark = localException != null ? ("executeLocalTransactionBranch exception: " + localException.toString()) : null;
130:
131: // 提交消息 COMMIT / ROLLBACK！！！通信方式为：Oneway！！！
132: this.mQClientFactory.getMQClientAPIImpl().endTransactionOneway(brokerAddr, requestHeader, remark, this.defaultMQProducer.getSendMsgTimeout());
133: }2.2 Broker 处理结束事务请求
- 查询请求的消息，进行提交 / 回滚。实现代码如下：
1: // 【EndTransactionProcessor.java】2: public RemotingCommand processRequest(ChannelHandlerContext ctx, RemotingCommand request) throws RemotingCommandException {3: final RemotingCommand response = RemotingCommand.createResponseCommand(null);4: final EndTransactionRequestHeader requestHeader = (EndTransactionRequestHeader) request.decodeCommandCustomHeader(EndTransactionRequestHeader.class);5: 6: // 省略代码 =》打印日志（只处理 COMMIT / ROLLBACK）7: 8: // 查询提交的消息9: final MessageExt msgExt = this.brokerController.getMessageStore().lookMessageByOffset(requestHeader.getCommitLogOffset());
10: if (msgExt != null) {
11: // 省略代码 =》校验消息
12:
13: // 生成消息
14: MessageExtBrokerInner msgInner = this.endMessageTransaction(msgExt);
15: msgInner.setSysFlag(MessageSysFlag.resetTransactionValue(msgInner.getSysFlag(), requestHeader.getCommitOrRollback()));
16: msgInner.setQueueOffset(requestHeader.getTranStateTableOffset());
17: msgInner.setPreparedTransactionOffset(requestHeader.getCommitLogOffset());
18: msgInner.setStoreTimestamp(msgExt.getStoreTimestamp());
19: if (MessageSysFlag.TRANSACTION_ROLLBACK_TYPE == requestHeader.getCommitOrRollback()) {
20: msgInner.setBody(null);
21: }
22:
23: // 存储生成消息
24: final MessageStore messageStore = this.brokerController.getMessageStore();
25: final PutMessageResult putMessageResult = messageStore.putMessage(msgInner);
26:
27: // 处理存储结果
28: if (putMessageResult != null) {
29: switch (putMessageResult.getPutMessageStatus()) {
30: // Success
31: case PUT_OK:
32: case FLUSH_DISK_TIMEOUT:
33: case FLUSH_SLAVE_TIMEOUT:
34: case SLAVE_NOT_AVAILABLE:
35: response.setCode(ResponseCode.SUCCESS);
36: response.setRemark(null);
37: break;
38: // Failed
39: case CREATE_MAPEDFILE_FAILED:
40: response.setCode(ResponseCode.SYSTEM_ERROR);
41: response.setRemark("create maped file failed.");
42: break;
43: case MESSAGE_ILLEGAL:
44: case PROPERTIES_SIZE_EXCEEDED:
45: response.setCode(ResponseCode.MESSAGE_ILLEGAL);
46: response.setRemark("the message is illegal, maybe msg body or properties length not matched. msg body length limit 128k, msg properties length limit 32k.");
47: break;
48: case SERVICE_NOT_AVAILABLE:
49: response.setCode(ResponseCode.SERVICE_NOT_AVAILABLE);
50: response.setRemark("service not available now.");
51: break;
52: case OS_PAGECACHE_BUSY:
53: response.setCode(ResponseCode.SYSTEM_ERROR);
54: response.setRemark("OS page cache busy, please try another machine");
55: break;
56: case UNKNOWN_ERROR:
57: response.setCode(ResponseCode.SYSTEM_ERROR);
58: response.setRemark("UNKNOWN_ERROR");
59: break;
60: default:
61: response.setCode(ResponseCode.SYSTEM_ERROR);
62: response.setRemark("UNKNOWN_ERROR DEFAULT");
63: break;
64: }
65:
66: return response;
67: } else {
68: response.setCode(ResponseCode.SYSTEM_ERROR);
69: response.setRemark("store putMessage return null");
70: }
71: } else {
72: response.setCode(ResponseCode.SYSTEM_ERROR);
73: response.setRemark("find prepared transaction message failed");
74: return response;
75: }
76:
77: return response;
78: }2.3 Broker 生成 ConsumeQueue
- 事务消息，提交（
COMMIT）后才生成ConsumeQueue。
1: // 【DefaultMessageStore.java】2: public void doDispatch(DispatchRequest req) {3: // 非事务消息 或 事务提交消息 建立 消息位置信息 到 ConsumeQueue4: final int tranType = MessageSysFlag.getTransactionValue(req.getSysFlag());5: switch (tranType) {6: case MessageSysFlag.TRANSACTION_NOT_TYPE: // 非事务消息7: case MessageSysFlag.TRANSACTION_COMMIT_TYPE: // 事务消息COMMIT8: DefaultMessageStore.this.putMessagePositionInfo(req.getTopic(), req.getQueueId(), req.getCommitLogOffset(), req.getMsgSize(),9: req.getTagsCode(), req.getStoreTimestamp(), req.getConsumeQueueOffset());
10: break;
11: case MessageSysFlag.TRANSACTION_PREPARED_TYPE: // 事务消息PREPARED
12: case MessageSysFlag.TRANSACTION_ROLLBACK_TYPE: // 事务消息ROLLBACK
13: break;
14: }
15: // 省略代码 => 建立 索引信息 到 IndexFile
16: }3. 事务消息回查
- 【事务消息回查】功能曾经开源过，目前（V4.0.0）暂未开源。如下是该功能的开源情况：
| 版本 | 【事务消息回查】 | |
|---|---|---|
| 官方V3.0.4 ~ V3.1.4 | 基于 文件系统 实现 | 已开源 |
| 官方V3.1.5 ~ V4.0.0 | 基于 数据库 实现 | 未完全开源 |
我们来看看两种情况下是怎么实现的。
3.1 Broker 发起【事务消息回查】
3.1.1 官方V3.1.4：基于文件系统
仓库地址：https://github.com/YunaiV/rocketmq-3.1.9/tree/release_3.1.4
相较于普通消息，【事务消息】多依赖如下三个组件：
- TransactionStateService ：事务状态服务，负责对【事务消息】进行管理，包括存储与更新事务消息状态、回查事务消息状态等等。
- TranStateTable ：【事务消息】状态存储。基于
MappedFileQueue实现，默认存储路径为~/store/transaction/statetable，每条【事务消息】状态存储结构如下：
| 第几位 | 字段 | 说明 | 数据类型 | 字节数 |
|---|---|---|---|---|
| 1 | offset | CommitLog 物理存储位置 | Long | 8 |
| 2 | size | 消息长度 | Int | 4 |
| 3 | timestamp | 消息存储时间，单位：秒 | Int | 4 |
| 4 | producerGroupHash | producerGroup 求 HashCode | Int | 4 |
| 5 | state | 事务状态 | Int | 4 |
- TranRedoLog ：
TranStateTable重放日志，每次写操作TranStateTable记录重放日志。当Broker异常关闭时，使用TranRedoLog恢复TranStateTable。基于ConsumeQueue实现，Topic为TRANSACTION_REDOLOG_TOPIC_XXXX，默认存储路径为~/store/transaction/redolog。
简单手绘逻辑图如下?：
3.1.1.1 存储消息到 CommitLog
- 存储【half消息】到
CommitLog时，消息队列位置（queueOffset）使用TranStateTable最大物理位置（可写入物理位置）。这样，消息可以索引到自己对应的TranStateTable的位置和记录。
核心代码如下：
1: // 【DefaultAppendMessageCallback.java】2: class DefaultAppendMessageCallback implements AppendMessageCallback {3: public AppendMessageResult doAppend(final long fileFromOffset, final ByteBuffer byteBuffer, final int maxBlank, final Object msg) {4: // ... ...5: 6: // 事务消息需要特殊处理 7: final int tranType = MessageSysFlag.getTransactionValue(msgInner.getSysFlag());8: switch (tranType) {9: case MessageSysFlag.TransactionPreparedType: // 消息队列位置（queueOffset）使用 TranStateTable 最大物理位置（可写入物理位置）
10: queueOffset = CommitLog.this.defaultMessageStore.getTransactionStateService().getTranStateTableOffset().get();
11: break;
12: case MessageSysFlag.TransactionRollbackType:
13: queueOffset = msgInner.getQueueOffset();
14: break;
15: case MessageSysFlag.TransactionNotType:
16: case MessageSysFlag.TransactionCommitType:
17: default:
18: break;
19: }
20:
21: // ... ...
22:
23: switch (tranType) {
24: case MessageSysFlag.TransactionPreparedType:
25: // 更新 TranStateTable 最大物理位置（可写入物理位置）
26: CommitLog.this.defaultMessageStore.getTransactionStateService().getTranStateTableOffset().incrementAndGet();
27: break;
28: case MessageSysFlag.TransactionRollbackType:
29: break;
30: case MessageSysFlag.TransactionNotType:
31: case MessageSysFlag.TransactionCommitType:
32: // 更新下一次的ConsumeQueue信息
33: CommitLog.this.topicQueueTable.put(key, ++queueOffset);
34: break;
35: default:
36: break;
37: }
38:
39: // 返回结果
40: return result;
41: }
42: }3.1.1.2 写【事务消息】状态存储（TranStateTable）
- 处理【Half消息】时，新增【事务消息】状态存储（
TranStateTable）。 - 处理【Commit / Rollback消息】时，更新 【事务消息】状态存储（
TranStateTable） COMMIT / ROLLBACK。 - 每次写操作【事务消息】状态存储（
TranStateTable），记录重放日志（TranRedoLog）。
核心代码如下：
1: // 【DispatchMessageService.java】2: private void doDispatch() {3: if (!this.requestsRead.isEmpty()) {4: for (DispatchRequest req : this.requestsRead) {5: 6: // ... ...7: 8: // 2、写【事务消息】状态存储（TranStateTable）9: if (req.getProducerGroup() != null) {10: switch (tranType) {11: case MessageSysFlag.TransactionNotType:12: break;13: case MessageSysFlag.TransactionPreparedType:14: // 新增 【事务消息】状态存储（TranStateTable）15: DefaultMessageStore.this.getTransactionStateService().appendPreparedTransaction(16: req.getCommitLogOffset(), req.getMsgSize(), (int) (req.getStoreTimestamp() / 1000), req.getProducerGroup().hashCode());17: break;18: case MessageSysFlag.TransactionCommitType:19: case MessageSysFlag.TransactionRollbackType:20: // 更新 【事务消息】状态存储（TranStateTable） COMMIT / ROLLBACK21: DefaultMessageStore.this.getTransactionStateService().updateTransactionState(22: req.getTranStateTableOffset(), req.getPreparedTransactionOffset(), req.getProducerGroup().hashCode(), tranType);23: break;24: }25: }26: // 3、记录 TranRedoLog27: switch (tranType) {28: case MessageSysFlag.TransactionNotType:29: break;30: case MessageSysFlag.TransactionPreparedType:31: // 记录 TranRedoLog32: DefaultMessageStore.this.getTransactionStateService().getTranRedoLog().putMessagePostionInfoWrapper(33: req.getCommitLogOffset(), req.getMsgSize(), TransactionStateService.PreparedMessageTagsCode,34: req.getStoreTimestamp(), 0L);35: break;36: case MessageSysFlag.TransactionCommitType:37: case MessageSysFlag.TransactionRollbackType:38: // 记录 TranRedoLog39: DefaultMessageStore.this.getTransactionStateService().getTranRedoLog().putMessagePostionInfoWrapper(40: req.getCommitLogOffset(), req.getMsgSize(), req.getPreparedTransactionOffset(),41: req.getStoreTimestamp(), 0L);42: break;43: }44: }45: 46: // ...省略代码47: }48: }49: // ⬇️⬇️⬇️【TransactionStateService.java】50: /**51: * 新增事务状态52: *53: * @param clOffset commitLog 物理位置54: * @param size 消息长度55: * @param timestamp 消息存储时间56: * @param groupHashCode groupHashCode57: * @return 是否成功58: */59: public boolean appendPreparedTransaction(//60: final long clOffset,//61: final int size,//62: final int timestamp,//63: final int groupHashCode//64: ) {65: MapedFile mapedFile = this.tranStateTable.getLastMapedFile();66: if (null == mapedFile) {67: log.error("appendPreparedTransaction: create mapedfile error.");68: return false;69: }70: 71: // 首次创建，加入定时任务中72: if (0 == mapedFile.getWrotePostion()) {73: this.addTimerTask(mapedFile);74: }75: 76: this.byteBufferAppend.position(0);77: this.byteBufferAppend.limit(TSStoreUnitSize);78: 79: // Commit Log Offset80: this.byteBufferAppend.putLong(clOffset);81: // Message Size82: this.byteBufferAppend.putInt(size);83: // Timestamp84: this.byteBufferAppend.putInt(timestamp);85: // Producer Group Hashcode86: this.byteBufferAppend.putInt(groupHashCode);87: // Transaction State88: this.byteBufferAppend.putInt(MessageSysFlag.TransactionPreparedType);89: 90: return mapedFile.appendMessage(this.byteBufferAppend.array());91: }92: 93: /**94: * 更新事务状态95: *96: * @param tsOffset tranStateTable 物理位置97: * @param clOffset commitLog 物理位置98: * @param groupHashCode groupHashCode99: * @param state 事务状态
100: * @return 是否成功
101: */
102: public boolean updateTransactionState(
103: final long tsOffset,
104: final long clOffset,
105: final int groupHashCode,
106: final int state) {
107: SelectMapedBufferResult selectMapedBufferResult = this.findTransactionBuffer(tsOffset);
108: if (selectMapedBufferResult != null) {
109: try {
110: // 校验是否能够更新
111: // .... ...
112:
113: // 更新事务状态
114: selectMapedBufferResult.getByteBuffer().putInt(TS_STATE_POS, state);
115: }
116: catch (Exception e) {
117: log.error("updateTransactionState exception", e);
118: }
119: finally {
120: selectMapedBufferResult.release();
121: }
122: }
123:
124: return false;
125: }3.1.1.3 【事务消息】回查
TranStateTable每个MappedFile都对应一个Timer。Timer固定周期（默认：60s）遍历MappedFile，查找【half消息】，向Producer发起【事务消息】回查请求。【事务消息】回查结果的逻辑不在此处进行，在 CommitLog dispatch时执行。
实现代码如下：
1: // 【TransactionStateService.java】2: /**3: * 初始化定时任务4: */5: private void initTimerTask() {6: //7: final List<MapedFile> mapedFiles = this.tranStateTable.getMapedFiles();8: for (MapedFile mf : mapedFiles) {9: this.addTimerTask(mf);10: }11: }12: 13: /**14: * 每个文件初始化定时任务15: * @param mf 文件16: */17: private void addTimerTask(final MapedFile mf) {18: this.timer.scheduleAtFixedRate(new TimerTask() {19: private final MapedFile mapedFile = mf;20: private final TransactionCheckExecuter transactionCheckExecuter = TransactionStateService.this.defaultMessageStore.getTransactionCheckExecuter();21: private final long checkTransactionMessageAtleastInterval = TransactionStateService.this.defaultMessageStore.getMessageStoreConfig()22: .getCheckTransactionMessageAtleastInterval();23: private final boolean slave = TransactionStateService.this.defaultMessageStore.getMessageStoreConfig().getBrokerRole() == BrokerRole.SLAVE;24: 25: @Override26: public void run() {27: // Slave不需要回查事务状态28: if (slave) {29: return;30: }31: // Check功能是否开启32: if (!TransactionStateService.this.defaultMessageStore.getMessageStoreConfig()33: .isCheckTransactionMessageEnable()) {34: return;35: }36: 37: try {38: SelectMapedBufferResult selectMapedBufferResult = mapedFile.selectMapedBuffer(0);39: if (selectMapedBufferResult != null) {40: long preparedMessageCountInThisMapedFile = 0; // 回查的【half消息】数量41: int i = 0;42: try {43: // 循环每条【事务消息】状态，对【half消息】进行回查44: for (; i < selectMapedBufferResult.getSize(); i += TSStoreUnitSize) {45: selectMapedBufferResult.getByteBuffer().position(i);46: 47: // Commit Log Offset48: long clOffset = selectMapedBufferResult.getByteBuffer().getLong();49: // Message Size50: int msgSize = selectMapedBufferResult.getByteBuffer().getInt();51: // Timestamp52: int timestamp = selectMapedBufferResult.getByteBuffer().getInt();53: // Producer Group Hashcode54: int groupHashCode = selectMapedBufferResult.getByteBuffer().getInt();55: // Transaction State56: int tranType = selectMapedBufferResult.getByteBuffer().getInt();57: 58: // 已经提交或者回滚的消息跳过59: if (tranType != MessageSysFlag.TransactionPreparedType) {60: continue;61: }62: 63: // 遇到时间不符合最小轮询间隔，终止64: long timestampLong = timestamp * 1000;65: long diff = System.currentTimeMillis() - timestampLong;66: if (diff < checkTransactionMessageAtleastInterval) {67: break;68: }69: 70: preparedMessageCountInThisMapedFile++;71: 72: // 回查Producer73: try {74: this.transactionCheckExecuter.gotoCheck(groupHashCode, getTranStateOffset(i), clOffset, msgSize);75: } catch (Exception e) {76: tranlog.warn("gotoCheck Exception", e);77: }78: }79: 80: // 无回查的【half消息】数量，且遍历完，则终止定时任务81: if (0 == preparedMessageCountInThisMapedFile //82: && i == mapedFile.getFileSize()) {83: tranlog.info("remove the transaction timer task, because no prepared message in this mapedfile[{}]", mapedFile.getFileName());84: this.cancel();85: }86: } finally {87: selectMapedBufferResult.release();88: }89: 90: tranlog.info("the transaction timer task execute over in this period, {} Prepared Message: {} Check Progress: {}/{}", mapedFile.getFileName(),//91: preparedMessageCountInThisMapedFile, i / TSStoreUnitSize, mapedFile.getFileSize() / TSStoreUnitSize);92: } else if (mapedFile.isFull()) {93: tranlog.info("the mapedfile[{}] maybe deleted, cancel check transaction timer task", mapedFile.getFileName());94: this.cancel();95: return;96: }97: } catch (Exception e) {98: log.error("check transaction timer task Exception", e);99: }
100: }
101:
102:
103: private long getTranStateOffset(final long currentIndex) {
104: long offset = (this.mapedFile.getFileFromOffset() + currentIndex) / TransactionStateService.TSStoreUnitSize;
105: return offset;
106: }
107: }, 1000 * 60, this.defaultMessageStore.getMessageStoreConfig().getCheckTransactionMessageTimerInterval());
108: }
109:
110: // 【DefaultTransactionCheckExecuter.java】
111: @Override
112: public void gotoCheck(int producerGroupHashCode, long tranStateTableOffset, long commitLogOffset,
113: int msgSize) {
114: // 第一步、查询Producer
115: final ClientChannelInfo clientChannelInfo = this.brokerController.getProducerManager().pickProducerChannelRandomly(producerGroupHashCode);
116: if (null == clientChannelInfo) {
117: log.warn("check a producer transaction state, but not find any channel of this group[{}]", producerGroupHashCode);
118: return;
119: }
120:
121: // 第二步、查询消息
122: SelectMapedBufferResult selectMapedBufferResult = this.brokerController.getMessageStore().selectOneMessageByOffset(commitLogOffset, msgSize);
123: if (null == selectMapedBufferResult) {
124: log.warn("check a producer transaction state, but not find message by commitLogOffset: {}, msgSize: ", commitLogOffset, msgSize);
125: return;
126: }
127:
128: // 第三步、向Producer发起请求
129: final CheckTransactionStateRequestHeader requestHeader = new CheckTransactionStateRequestHeader();
130: requestHeader.setCommitLogOffset(commitLogOffset);
131: requestHeader.setTranStateTableOffset(tranStateTableOffset);
132: this.brokerController.getBroker2Client().checkProducerTransactionState(clientChannelInfo.getChannel(), requestHeader, selectMapedBufferResult);
133: }3.1.1.4 初始化【事务消息】状态存储（TranStateTable）
- 根据最后 Broker 关闭是否正常，会有不同的初始化方式。
核心代码如下：
1: // 【TransactionStateService.java】2: /**3: * 初始化 TranRedoLog4: * @param lastExitOK 是否正常退出5: */6: public void recoverStateTable(final boolean lastExitOK) {7: if (lastExitOK) {8: this.recoverStateTableNormal();9: } else {10: // 第一步，删除State Table11: this.tranStateTable.destroy();12: // 第二步，通过RedoLog全量恢复StateTable13: this.recreateStateTable();14: }15: }16: 17: /**18: * 扫描 TranRedoLog 重建 StateTable19: */20: private void recreateStateTable() {21: this.tranStateTable = new MapedFileQueue(StorePathConfigHelper.getTranStateTableStorePath(defaultMessageStore22: .getMessageStoreConfig().getStorePathRootDir()), defaultMessageStore23: .getMessageStoreConfig().getTranStateTableMapedFileSize(), null);24: 25: final TreeSet<Long> preparedItemSet = new TreeSet<Long>();26: 27: // 第一步，从头扫描RedoLog28: final long minOffset = this.tranRedoLog.getMinOffsetInQuque();29: long processOffset = minOffset;30: while (true) {31: SelectMapedBufferResult bufferConsumeQueue = this.tranRedoLog.getIndexBuffer(processOffset);32: if (bufferConsumeQueue != null) {33: try {34: long i = 0;35: for (; i < bufferConsumeQueue.getSize(); i += ConsumeQueue.CQStoreUnitSize) {36: long offsetMsg = bufferConsumeQueue.getByteBuffer().getLong();37: int sizeMsg = bufferConsumeQueue.getByteBuffer().getInt();38: long tagsCode = bufferConsumeQueue.getByteBuffer().getLong();39: 40: if (TransactionStateService.PreparedMessageTagsCode == tagsCode) { // Prepared41: preparedItemSet.add(offsetMsg);42: } else { // Commit/Rollback43: preparedItemSet.remove(tagsCode);44: }45: }46: 47: processOffset += i;48: } finally { // 必须释放资源49: bufferConsumeQueue.release();50: }51: } else {52: break;53: }54: }55: log.info("scan transaction redolog over, End offset: {}, Prepared Transaction Count: {}", processOffset, preparedItemSet.size());56: 57: // 第二步，重建StateTable58: Iterator<Long> it = preparedItemSet.iterator();59: while (it.hasNext()) {60: Long offset = it.next();61: MessageExt msgExt = this.defaultMessageStore.lookMessageByOffset(offset);62: if (msgExt != null) {63: this.appendPreparedTransaction(msgExt.getCommitLogOffset(), msgExt.getStoreSize(),64: (int) (msgExt.getStoreTimestamp() / 1000),65: msgExt.getProperty(MessageConst.PROPERTY_PRODUCER_GROUP).hashCode());66: this.tranStateTableOffset.incrementAndGet();67: }68: }69: }70: 71: /**72: * 加载（解析）TranStateTable 的 MappedFile73: * 1. 清理多余 MappedFile，设置最后一个 MappedFile的写入位置(position74: * 2. 设置 TanStateTable 最大物理位置（可写入位置）75: */76: private void recoverStateTableNormal() {77: final List<MapedFile> mapedFiles = this.tranStateTable.getMapedFiles();78: if (!mapedFiles.isEmpty()) {79: // 从倒数第三个文件开始恢复80: int index = mapedFiles.size() - 3;81: if (index < 0) {82: index = 0;83: }84: 85: int mapedFileSizeLogics = this.tranStateTable.getMapedFileSize();86: MapedFile mapedFile = mapedFiles.get(index);87: ByteBuffer byteBuffer = mapedFile.sliceByteBuffer();88: long processOffset = mapedFile.getFileFromOffset();89: long mapedFileOffset = 0;90: while (true) {91: for (int i = 0; i < mapedFileSizeLogics; i += TSStoreUnitSize) {92: 93: final long clOffset_read = byteBuffer.getLong();94: final int size_read = byteBuffer.getInt();95: final int timestamp_read = byteBuffer.getInt();96: final int groupHashCode_read = byteBuffer.getInt();97: final int state_read = byteBuffer.getInt();98: 99: boolean stateOK = false;
100: switch (state_read) {
101: case MessageSysFlag.TransactionPreparedType:
102: case MessageSysFlag.TransactionCommitType:
103: case MessageSysFlag.TransactionRollbackType:
104: stateOK = true;
105: break;
106: default:
107: break;
108: }
109:
110: // 说明当前存储单元有效
111: if (clOffset_read >= 0 && size_read > 0 && stateOK) {
112: mapedFileOffset = i + TSStoreUnitSize;
113: } else {
114: log.info("recover current transaction state table file over, " + mapedFile.getFileName() + " "
115: + clOffset_read + " " + size_read + " " + timestamp_read);
116: break;
117: }
118: }
119:
120: // 走到文件末尾，切换至下一个文件
121: if (mapedFileOffset == mapedFileSizeLogics) {
122: index++;
123: if (index >= mapedFiles.size()) { // 循环while结束
124: log.info("recover last transaction state table file over, last maped file " + mapedFile.getFileName());
125: break;
126: } else { // 切换下一个文件
127: mapedFile = mapedFiles.get(index);
128: byteBuffer = mapedFile.sliceByteBuffer();
129: processOffset = mapedFile.getFileFromOffset();
130: mapedFileOffset = 0;
131: log.info("recover next transaction state table file, " + mapedFile.getFileName());
132: }
133: } else {
134: log.info("recover current transaction state table queue over " + mapedFile.getFileName() + " " + (processOffset + mapedFileOffset));
135: break;
136: }
137: }
138:
139: // 清理多余 MappedFile，设置最后一个 MappedFile的写入位置(position
140: processOffset += mapedFileOffset;
141: this.tranStateTable.truncateDirtyFiles(processOffset);
142:
143: // 设置 TanStateTable 最大物理位置（可写入位置）
144: this.tranStateTableOffset.set(this.tranStateTable.getMaxOffset() / TSStoreUnitSize);
145: log.info("recover normal over, transaction state table max offset: {}", this.tranStateTableOffset.get());
146: }
147: }3.1.1.5 补充
- 为什么 V3.1.5 开始，使用 数据库 实现【事务状态】的存储？如下是来自官方文档的说明，可能是一部分原因：
RocketMQ 这种实现事务方式，没有通过 KV 存储做，而是通过 Offset 方式，存在一个显著缺陷，即通过 Offset 更改数据，会令系统的脏页过多，需要特别关注。
3.1.2 官方V4.0.0：基于数据库
仓库地址：https://github.com/apache/incubator-rocketmq
官方V4.0.0 暂时未完全开源【事务消息回查】功能，So 我们需要进行一些猜想，可能不一定正确?。
?我们来对比【官方V3.1.4：基于文件】的实现。
- TransactionRecord ：记录每条【事务消息】。类似
TranStateTable。
| TranStateTable | TransactionRecord | |
|---|---|---|
| offset | offset | |
| producerGroupHash | producerGroup | |
| size | 无 | 非必须字段：【事务消息】回查时，使用 offset 读取 CommitLog 获得。 |
| timestamp | 无 | 非必须字段：【事务消息】回查时，使用 offset 读取 CommitLog 获得。 |
| state | 无 | 非必须字段： 事务开始，增加记录；事务结束，删除记录。 |
另外，数据库本身保证了数据存储的可靠性，无需 TranRedoLog。
简单手绘逻辑图如下?：
3.2 Producer 接收【事务消息回查】
- 顺序图如下：
- 核心代码如下：
1: // 【DefaultMQProducerImpl.java】2: /**3: * 检查【事务状态】状态4: *5: * @param addr broker地址6: * @param msg 消息7: * @param header 请求8: */9: @Override10: public void checkTransactionState(final String addr, final MessageExt msg, final CheckTransactionStateRequestHeader header) {11: Runnable request = new Runnable() {12: private final String brokerAddr = addr;13: private final MessageExt message = msg;14: private final CheckTransactionStateRequestHeader checkRequestHeader = header;15: private final String group = DefaultMQProducerImpl.this.defaultMQProducer.getProducerGroup();16: 17: @Override18: public void run() {19: TransactionCheckListener transactionCheckListener = DefaultMQProducerImpl.this.checkListener();20: if (transactionCheckListener != null) {21: // 获取事务执行状态22: LocalTransactionState localTransactionState = LocalTransactionState.UNKNOW;23: Throwable exception = null;24: try {25: localTransactionState = transactionCheckListener.checkLocalTransactionState(message);26: } catch (Throwable e) {27: log.error("Broker call checkTransactionState, but checkLocalTransactionState exception", e);28: exception = e;29: }30: 31: // 处理事务结果，提交消息 COMMIT / ROLLBACK32: this.processTransactionState(//33: localTransactionState, //34: group, //35: exception);36: } else {37: log.warn("checkTransactionState, pick transactionCheckListener by group[{}] failed", group);38: }39: }40: 41: /**42: * 处理事务结果，提交消息 COMMIT / ROLLBACK43: *44: * @param localTransactionState 【本地事务】状态45: * @param producerGroup producerGroup46: * @param exception 检查【本地事务】状态发生的异常47: */48: private void processTransactionState(//49: final LocalTransactionState localTransactionState, //50: final String producerGroup, //51: final Throwable exception) {52: final EndTransactionRequestHeader thisHeader = new EndTransactionRequestHeader();53: thisHeader.setCommitLogOffset(checkRequestHeader.getCommitLogOffset());54: thisHeader.setProducerGroup(producerGroup);55: thisHeader.setTranStateTableOffset(checkRequestHeader.getTranStateTableOffset());56: thisHeader.setFromTransactionCheck(true);57: 58: // 设置消息编号59: String uniqueKey = message.getProperties().get(MessageConst.PROPERTY_UNIQ_CLIENT_MESSAGE_ID_KEYIDX);60: if (uniqueKey == null) {61: uniqueKey = message.getMsgId();62: }63: thisHeader.setMsgId(uniqueKey);64: 65: thisHeader.setTransactionId(checkRequestHeader.getTransactionId());66: switch (localTransactionState) {67: case COMMIT_MESSAGE:68: thisHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_COMMIT_TYPE);69: break;70: case ROLLBACK_MESSAGE:71: thisHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_ROLLBACK_TYPE);72: log.warn("when broker check, client rollback this transaction, {}", thisHeader);73: break;74: case UNKNOW:75: thisHeader.setCommitOrRollback(MessageSysFlag.TRANSACTION_NOT_TYPE);76: log.warn("when broker check, client does not know this transaction state, {}", thisHeader);77: break;78: default:79: break;80: }81: 82: String remark = null;83: if (exception != null) {84: remark = "checkLocalTransactionState Exception: " + RemotingHelper.exceptionSimpleDesc(exception);85: }86: 87: try {88: // 提交消息 COMMIT / ROLLBACK89: DefaultMQProducerImpl.this.mQClientFactory.getMQClientAPIImpl().endTransactionOneway(brokerAddr, thisHeader, remark,90: 3000);91: } catch (Exception e) {92: log.error("endTransactionOneway exception", e);93: }94: }95: };96: 97: // 提交执行98: this.checkExecutor.submit(request);99: }
100:
101: // 【DefaultMQProducerImpl.java】
102: /**
103: * 【事务消息回查】检查监听器
104: */
105: public interface TransactionCheckListener {
106:
107: /**
108: * 获取（检查）【本地事务】状态
109: *
110: * @param msg 消息
111: * @return 事务状态
112: */
113: LocalTransactionState checkLocalTransactionState(final MessageExt msg);
114:
115: }转载于:https://my.oschina.net/oosc/blog/1795277
RocketMQ 源码分析 事务消息相关推荐
- RocketMQ源码解析-事务消息的二阶段提交
在生产者producer当中,通过sendMessageInTransaction()方法来发送事务消息,但是在一开始向Broker发送的事务消息的时候,具体的事务操作还并没有进行处理,而是相当于向B ...
- RocketMQ源码分析之延迟消息
文章目录 前言 一.延迟消息 1.特点 2.使用场景 3.demo 二.发送延迟消息 三.broker端存储延迟消息 四.总结 1.延迟消息工作原理 2.延迟消息在消费者消费重试中的应用 前言 本篇文 ...
- 《RocketMQ源码分析》NameServer如何处理Broker的连接
<RocketMQ源码分析>NameServer如何处理Broker的连接 NameServer 介绍 NameServer 功能 动态路由发现和注册 服务剔除 创建NameServerC ...
- 9. 源码分析之消息消费
源码分析之消息消费 Rebalance(针对集群消费模式) (1)消费Group下的所有消费者 (2)Topic的所有Queue队列 (3)Queue分配策略 触发时机 (1)消费者启动 (2)消费者 ...
- 事务回滚什么意思 try_分布式事务 TCC-Transaction 源码分析——事务恢复
1. 概述 本文分享 TCC 恢复.主要涉及如下二个 package 路径下的类: org.mengyun.tcctransaction.recover RecoverConfig,事务恢复配置接口 ...
- RocketMQ源码(十)—Broker 消息刷盘服务GroupCommitService、FlushRealTimeService、CommitRealTimeService源码深度解析
深入的介绍了broker的消息刷盘服务源码解析,以及高性能的刷盘机制. 学习RocketMQ的时候,我们知道RocketMQ的刷盘策略有两个,同步或者是异步: 1. 同步刷盘:如上图所示,只有消息真正 ...
- RocketMQ源码解析之消息消费者(consume Message)
原创不易,转载请注明出处 文章目录 前言 1.消息流程介绍 2.源码解析 2.1 并发消费 2.2 顺序消费 前言 我们在<RocketMQ源码解析之消息消费者(pullMessage)> ...
- RocketMQ源码分析之RocketMQ事务消息实现原理上篇(二阶段提交)
在阅读本文前,若您对RocketMQ技术感兴趣,请加入 RocketMQ技术交流群 根据上文的描述,发送事务消息的入口为: TransactionMQProducer#sendMessageInTra ...
- RocketMQ源码分析之从官方示例窥探：RocketMQ事务消息实现基本思想
RocketMQ4.3.0版本开始支持事务消息,后续分享将开始将剖析事务消息的实现原理.首先从官方给出的Demo实例入手,以此通往RocketMQ事务消息的世界中. 官方版本未发布之前,从apache ...
最新文章
- Android之View绘制流程源码分析
- C_functions
- 10-20国际标准导联系统
- boost::hana::string_c用法的测试程序
- 前端学习（501）：水平居中布局得第二种方式的优缺点
- linux的基础知识——时序竞态
- spring @value
- SLF4J 与Log4J
- Q95：纹理映射（Texture Mapping）（2）——圆柱面
- css3优惠卷上方锯齿_CSS不规则卡片，纯CSS制作优惠券样式，CSS实现锯齿样式
- php网页代码字体大小,HTML网页字体大小的设置
- uni 加入图标_uniapp如何更改图标
- 一个div实现太极图案+动画（简单易懂）
- 看不见的竞争 之 把握意见领袖
- [Windows] 迅雷 无修改 无限制 无视封锁
- 解决方案 word excel vba 以下隐藏模块存在编译错误
- java 企业邮箱_java 企业邮箱群发邮件
- rabbitmq消息发布mandatory参数
- 小程序倒计时轮播图、导航切换轮播
- Banana Pi R2 Pro使用EC25 4G模块实现路由功能，萤石云摄像头配置。
热门文章
- oracle(thin),kettle thin方式连接oracle,MySQL、SqlServer
- centos7不识别固态硬盘_固态硬盘分区与不分区哪个好【详细介绍】
- 矩阵的二范数_Python Numpy中的范数
- upload-labs_pass14_pass15_pass16_图片马的制作_结合文件包含漏洞_getimagesize_exif_imagetype
- 20220130---CTF WEB方向---命令执行和URL解码
- 基于matlab的语音信号基本处理系统,基于matlab的语音信号处理及分析
- JS获取标签方法及兼容处理
- 为什么要用webUI?
- CentOS 6.6 安装pip
- luogu P1058 立体图