在【RocketMQ】訊息的拉取一文中可知,消費者在啟動的時候,會建立訊息拉取API物件PullAPIWrapper
,呼叫pullKernelImpl方法向Broker傳送拉取訊息的請求,那麼在主從模式下消費者是如何選擇向哪個Broker傳送拉取請求的?
進入pullKernelImpl方法中,可以看到會呼叫recalculatePullFromWhichNode方法選擇一個Broker:
public class PullAPIWrapper {
public PullResult pullKernelImpl(
final MessageQueue mq,
final String subExpression,
final String expressionType,
final long subVersion,
final long offset,
final int maxNums,
final int sysFlag,
final long commitOffset,
final long brokerSuspendMaxTimeMillis,
final long timeoutMillis,
final CommunicationMode communicationMode,
final PullCallback pullCallback
) throws MQClientException, RemotingException, MQBrokerException, InterruptedException {
// 呼叫recalculatePullFromWhichNode方法獲取Broker ID,再呼叫findBrokerAddressInSubscribe根據ID獲取Broker的相關資訊
FindBrokerResult findBrokerResult =
this.mQClientFactory.findBrokerAddressInSubscribe(mq.getBrokerName(),
this.recalculatePullFromWhichNode(mq), false);
// ...
if (findBrokerResult != null) {
// ...
// 獲取Broker地址
String brokerAddr = findBrokerResult.getBrokerAddr();
if (PullSysFlag.hasClassFilterFlag(sysFlagInner)) {
brokerAddr = computePullFromWhichFilterServer(mq.getTopic(), brokerAddr);
}
// 傳送訊息拉取請求
PullResult pullResult = this.mQClientFactory.getMQClientAPIImpl().pullMessage(
brokerAddr,
requestHeader,
timeoutMillis,
communicationMode,
pullCallback);
return pullResult;
}
}
}
在recalculatePullFromWhichNode
方法中,會從pullFromWhichNodeTable
中根據訊息佇列獲取一個建議的Broker ID,如果獲取為空就返回Master節點的Broker ID,ROCKETMQ中Master角色的Broker ID為0,既然從pullFromWhichNodeTable
中可以知道從哪個Broker拉取資料,那麼pullFromWhichNodeTable
中的資料又是從哪裡來的?
public class PullAPIWrapper {
// KEY為訊息佇列,VALUE為建議的Broker ID
private ConcurrentMap<MessageQueue, AtomicLong/* brokerId */> pullFromWhichNodeTable =
new ConcurrentHashMap<MessageQueue, AtomicLong>(32);
public long recalculatePullFromWhichNode(final MessageQueue mq) {
if (this.isConnectBrokerByUser()) {
return this.defaultBrokerId;
}
// 從pullFromWhichNodeTable中獲取建議的broker ID
AtomicLong suggest = this.pullFromWhichNodeTable.get(mq);
if (suggest != null) {
return suggest.get();
}
// 返回Master Broker ID
return MixAll.MASTER_ID;
}
}
透過呼叫關係可知,在updatePullFromWhichNode
方法中更新了pullFromWhichNodeTable
的值,而updatePullFromWhichNode
方法又是被processPullResult
方法呼叫的,消費者向Broker傳送拉取訊息請求後,Broker對拉取請求進行處理時會設定一個broker ID(後面會講到),建議下次從這個Broker拉取訊息,消費者對拉取請求返回的響應資料進行處理時會呼叫processPullResult
方法,在這裡將建議的BrokerID取出,呼叫updatePullFromWhichNode
方法將其加入到了pullFromWhichNodeTable
中:
public class PullAPIWrapper {
private ConcurrentMap<MessageQueue, AtomicLong/* brokerId */> pullFromWhichNodeTable =
new ConcurrentHashMap<MessageQueue, AtomicLong>(32);
public PullResult processPullResult(final MessageQueue mq, final PullResult pullResult,
final SubscriptionData subscriptionData) {
PullResultExt pullResultExt = (PullResultExt) pullResult;
// 將拉取訊息請求返回的建議Broker ID,加入到pullFromWhichNodeTable中
this.updatePullFromWhichNode(mq, pullResultExt.getSuggestWhichBrokerId());
// ...
}
public void updatePullFromWhichNode(final MessageQueue mq, final long brokerId) {
AtomicLong suggest = this.pullFromWhichNodeTable.get(mq);
if (null == suggest) {
// 向pullFromWhichNodeTable中新增資料
this.pullFromWhichNodeTable.put(mq, new AtomicLong(brokerId));
} else {
suggest.set(brokerId);
}
}
}
接下來去看下是根據什麼條件決定選擇哪個Broker的。
返回建議的BrokerID
Broker在處理消費者拉取請求時,會呼叫PullMessageProcessor
的processRequest
方法,首先會呼叫MessageStore
的getMessage方法獲取訊息內容,在返回的結果GetMessageResult
中設定了一個是否建議從Slave節點拉取的屬性(這個值的設定稍後再說),會根據是否建議從slave節點進行以下處理:
- 如果建議從slave節點拉取訊息,會呼叫
subscriptionGroupConfig
訂閱分組配置的getWhichBrokerWhenConsumeSlowly
方法獲取從節點將ID設定到響應中,否則下次依舊建議從主節點拉取訊息,將MASTER節點的ID設定到響應中; - 判斷當前Broker的角色,如果是slave節點,並且配置了不允許從slave節點讀取資料(SlaveReadEnable = false),此時依舊建議從主節點拉取訊息,將MASTER節點的ID設定到響應中;
- 如果開啟了允許從slave節點讀取資料(SlaveReadEnable = true),有以下兩種情況:
- 如果建議從slave節點拉訊息,從訂閱分組配置中獲取從節點的ID,將ID設定到響應中;
- 如果不建議從slave節點拉取訊息,從訂閱分組配置中獲取設定的Broker Id;
當然,如果未開啟允許從Slave節點讀取資料,下次依舊建議從Master節點拉取;
訂閱分組配置
mqadmin命令的-i
引數可以指定從哪個Broker消費訊息(subscriptionGroupConfig
的getBrokerId
返回的值),-w
引數可以指定建議從slave節點消費的時候,從哪個slave消費(subscriptionGroupConfig
的getWhichBrokerWhenConsumeSlowly
方法返回的值):
usage: mqadmin updateSubGroup [-a <arg>] [-b <arg>] [-c <arg>] [-d <arg>] -g <arg> [-h] [-i <arg>] [-m <arg>]
[-n <arg>] [-q <arg>] [-r <arg>] [-s <arg>] [-w <arg>]
-i,--brokerId <arg> consumer from which broker id
-w,--whichBrokerWhenConsumeSlowly <arg> which broker id when consume slowly
public class PullMessageProcessor extends AsyncNettyRequestProcessor implements NettyRequestProcessor {
private RemotingCommand processRequest(final Channel channel, RemotingCommand request, boolean brokerAllowSuspend)
throws RemotingCommandException {
// ...
// 根據拉取偏移量獲取訊息
final GetMessageResult getMessageResult =
this.brokerController.getMessageStore().getMessage(requestHeader.getConsumerGroup(), requestHeader.getTopic(),
requestHeader.getQueueId(), requestHeader.getQueueOffset(), requestHeader.getMaxMsgNums(), messageFilter);
if (getMessageResult != null) {
response.setRemark(getMessageResult.getStatus().name());
responseHeader.setNextBeginOffset(getMessageResult.getNextBeginOffset());
responseHeader.setMinOffset(getMessageResult.getMinOffset());
responseHeader.setMaxOffset(getMessageResult.getMaxOffset());
// 是否建議從從節點拉取訊息
if (getMessageResult.isSuggestPullingFromSlave()) {
// 選擇一個從節點
responseHeader.setSuggestWhichBrokerId(subscriptionGroupConfig.getWhichBrokerWhenConsumeSlowly());
} else {
responseHeader.setSuggestWhichBrokerId(MixAll.MASTER_ID);
}
// 判斷Broker的角色
switch (this.brokerController.getMessageStoreConfig().getBrokerRole()) {
case ASYNC_MASTER:
case SYNC_MASTER:
break;
case SLAVE:
// 如果不允許從從節點讀取資料,設定為MasterID
if (!this.brokerController.getBrokerConfig().isSlaveReadEnable()) {
response.setCode(ResponseCode.PULL_RETRY_IMMEDIATELY);
responseHeader.setSuggestWhichBrokerId(MixAll.MASTER_ID);
}
break;
}
// 如果開啟了允許從從節點讀取資料
if (this.brokerController.getBrokerConfig().isSlaveReadEnable()) {
// 如果建議從從節點拉訊息
if (getMessageResult.isSuggestPullingFromSlave()) {
// 獲取從節點
responseHeader.setSuggestWhichBrokerId(subscriptionGroupConfig.getWhichBrokerWhenConsumeSlowly());
}
else {
// 獲取指定的broker
responseHeader.setSuggestWhichBrokerId(subscriptionGroupConfig.getBrokerId());
}
} else {
// 使用Master節點
responseHeader.setSuggestWhichBrokerId(MixAll.MASTER_ID);
}
} else {
response.setCode(ResponseCode.SYSTEM_ERROR);
response.setRemark("store getMessage return null");
}
}
}
是否建議從Slave節點拉取的設定
DefaultMessageStore的getMessage方法中用於獲取訊息內容,並會根據消費者的拉取進度判斷是否建議下次從Slave節點拉取訊息,判斷過程如下:
- diff:當前CommitLog最大的偏移量減去本次拉取訊息的最大物理偏移量,表示剩餘未拉取的訊息;
- memory:訊息在PageCache中的總大小,計算方式是總實體記憶體 * 訊息儲存在記憶體中的閥值(預設為40)/100,也就是說MQ會快取一部分訊息在作業系統的PageCache中,加速訪問;
- 如果diff大於memory,表示未拉取的訊息過多,已經超出了PageCache快取的資料的大小,還需要從磁碟中獲取訊息,所以此時會建議下次從Slave節點拉取;
public class DefaultMessageStore implements MessageStore {
public GetMessageResult getMessage(final String group, final String topic, final int queueId, final long offset,
final int maxMsgNums,
final MessageFilter messageFilter) {
// ...
// 當前CommitLog的最大偏移量
final long maxOffsetPy = this.commitLog.getMaxOffset();
ConsumeQueue consumeQueue = findConsumeQueue(topic, queueId);
if (consumeQueue != null) {
minOffset = consumeQueue.getMinOffsetInQueue();
maxOffset = consumeQueue.getMaxOffsetInQueue();
if (maxOffset == 0) {
// ...
} else {
// 根據消費進度獲取訊息佇列
SelectMappedBufferResult bufferConsumeQueue = consumeQueue.getIndexBuffer(offset);
if (bufferConsumeQueue != null) {
try {
// ...
// CommitLog最大偏移量減去本次拉取訊息的最大物理偏移量
long diff = maxOffsetPy - maxPhyOffsetPulling;
// 計算訊息在PageCache中的總大小(總實體記憶體 * 訊息儲存在記憶體中的閥值/100)
long memory = (long) (StoreUtil.TOTAL_PHYSICAL_MEMORY_SIZE
* (this.messageStoreConfig.getAccessMessageInMemoryMaxRatio() / 100.0));
// 是否建議下次去從節點拉取訊息
getResult.setSuggestPullingFromSlave(diff > memory);
} finally {
bufferConsumeQueue.release();
}
} else {
// ...
}
}
} else {
status = GetMessageStatus.NO_MATCHED_LOGIC_QUEUE;
nextBeginOffset = nextOffsetCorrection(offset, 0);
}
// ...
return getResult;
}
}
總結
消費者在啟動後需要向Broker傳送拉取訊息的請求,Broker收到請求後會根據訊息的拉取進度,返回一個建議的BrokerID,並設定到響應中返回,消費者處理響應時將建議的BrokerID放入pullFromWhichNodeTable,下次拉去訊息的時候從pullFromWhichNodeTable中取出,並向其傳送請求拉取訊息。
消費進度持久化
上面講解了主從模式下如何選擇從哪個Broker拉取訊息,接下來看下消費進度的持久化,因為廣播模式下消費進度儲存在每個消費者端,叢集模式下消費進度儲存在Broker端,所以接下來以叢集模式為例。
在【RocketMQ】訊息的拉取一文中可知,叢集模式下主要是透過RemoteBrokerOffsetStore
進行消費進度管理的,在持久化方法persistAll
中會呼叫updateConsumeOffsetToBroker
更新Broker端的消費進度:
public class RemoteBrokerOffsetStore implements OffsetStore {
@Override
public void persistAll(Set<MessageQueue> mqs) {
if (null == mqs || mqs.isEmpty())
return;
final HashSet<MessageQueue> unusedMQ = new HashSet<MessageQueue>();
for (Map.Entry<MessageQueue, AtomicLong> entry : this.offsetTable.entrySet()) {
MessageQueue mq = entry.getKey();
AtomicLong offset = entry.getValue();
if (offset != null) {
if (mqs.contains(mq)) {
try {
// 向Broker傳送請求更新消費進度
this.updateConsumeOffsetToBroker(mq, offset.get());
log.info("[persistAll] Group: {} ClientId: {} updateConsumeOffsetToBroker {} {}",
this.groupName,
this.mQClientFactory.getClientId(),
mq,
offset.get());
} catch (Exception e) {
log.error("updateConsumeOffsetToBroker exception, " + mq.toString(), e);
}
} else {
unusedMQ.add(mq);
}
}
}
// ...
}
}
由於updateConsumeOffsetToBroker
方法中先呼叫了findBrokerAddressInSubscribe
方法獲取Broker的資訊,所以這裡先看findBrokerAddressInSubscribe
方法是如何選擇Broker的,它需要傳入三個引數,分別為:Broker名稱、Broker ID、是否只查詢引數中傳入的那個BrokerID,方法的處理邏輯如下:
- 首先從
brokerAddrTable
中根據Broker的名稱獲取所有的Broker集合(主從模式下他們的Broker名稱一致,但是ID不一致),KEY為BrokerID,VALUE為Broker的地址; - 從Broker集合中根據引數中傳入的ID獲取broker地址;
- 判斷引數中傳入的BrokerID是否是主節點,記錄在slave變數中;
- 判斷獲取的Broker地址是否為空,記錄在found變數中;
- 如果根據BrokerId獲取的地址為空並且引數中傳入的BrokerId為從節點,繼續輪詢獲取下一個Broker,並判斷地址是否為空;
- 如果此時地址依舊為空並且onlyThisBroker傳入的false(也就是不必須選擇引數中傳入的那個BrokerID),此時獲取map集合中的第一個節點;
- 判斷獲取到的Broker地址是否為空,不為空封裝結果返回,否則返回NULL;
public class MQClientInstance {
public FindBrokerResult findBrokerAddressInSubscribe(
final String brokerName, // Broker名稱
final long brokerId, // Broker ID
final boolean onlyThisBroker // 是否只查詢引數中傳入的那個BrokerID
) {
String brokerAddr = null;
boolean slave = false;
boolean found = false;
// 獲取所有的Broker ID
HashMap<Long/* brokerId */, String/* address */> map = this.brokerAddrTable.get(brokerName);
if (map != null && !map.isEmpty()) {
brokerAddr = map.get(brokerId);
// 是否是從節點
slave = brokerId != MixAll.MASTER_ID;
// 地址是否為空
found = brokerAddr != null;
// 如果地址為空並且是從節點
if (!found && slave) {
// 獲取下一個Broker
brokerAddr = map.get(brokerId + 1);
found = brokerAddr != null;
}
// 如果地址為空
if (!found && !onlyThisBroker) {
// 獲取集合中的第一個節點
Entry<Long, String> entry = map.entrySet().iterator().next();
// 獲取地址
brokerAddr = entry.getValue();
// 是否是從節點
slave = entry.getKey() != MixAll.MASTER_ID;
// 置為true
found = true;
}
}
if (found) {
// 返回資料
return new FindBrokerResult(brokerAddr, slave, findBrokerVersion(brokerName, brokerAddr));
}
return null;
}
}
回到updateConsumeOffsetToBroker方法,先看第一次呼叫findBrokerAddressInSubscribe方法獲取Broker資訊,傳入的三個引數分別為:Broker名稱、Master節點的ID、true,根據上面講解的findBrokerAddressInSubscribe
方法裡面的查詢邏輯,如果查詢到Master節點的資訊,就正常返回,如果此時Master當機未能正常查詢到,由於傳入的Master節點的ID並且onlyThisBroker置為true,所以會查詢失敗返回NULL。
如果第一次呼叫為空,會進行第二次呼叫,與第一次呼叫不同的地方是第三個引數置為了false,也就是說不是必須選擇引數中指定的那個Broker,此時依舊優先查詢Master節點,如果Master節點未查詢到,由於onlyThisBroker置為了false,會迭代集合選擇第一個節點返回,此時返回的有可能是從節點。
總結:消費者會優先選擇向主節點傳送請求進行消費進度儲存,假如主節點當機等原因未能獲取到主節點的資訊,會迭代集合選擇第一個節點返回,所以消費者也可以向從節點傳送請求進行進度儲存,待主節點恢復後,依舊優先選擇主節點。
public class RemoteBrokerOffsetStore implements OffsetStore {
private void updateConsumeOffsetToBroker(MessageQueue mq, long offset) throws RemotingException,
MQBrokerException, InterruptedException, MQClientException {
// 更新消費進度
updateConsumeOffsetToBroker(mq, offset, true);
}
@Override
public void updateConsumeOffsetToBroker(MessageQueue mq, long offset, boolean isOneway) throws RemotingException,
MQBrokerException, InterruptedException, MQClientException {
// 第一次呼叫findBrokerAddressInSubscribe方法獲取Broker資訊,三個引數分別為:Broker名稱、Master節點的ID、true
FindBrokerResult findBrokerResult = this.mQClientFactory.findBrokerAddressInSubscribe(mq.getBrokerName(), MixAll.MASTER_ID, true);
// 如果獲取為空,進行第二次呼叫
if (null == findBrokerResult) {
// 三個引數分別為:Broker名稱、Master節點的ID、false
this.mQClientFactory.updateTopicRouteInfoFromNameServer(mq.getTopic());
findBrokerResult = this.mQClientFactory.findBrokerAddressInSubscribe(mq.getBrokerName(), MixAll.MASTER_ID, false);
}
if (findBrokerResult != null) {
// 設定請求頭
UpdateConsumerOffsetRequestHeader requestHeader = new UpdateConsumerOffsetRequestHeader();
requestHeader.setTopic(mq.getTopic());
requestHeader.setConsumerGroup(this.groupName);
requestHeader.setQueueId(mq.getQueueId());
requestHeader.setCommitOffset(offset);
// 傳送儲存消費進度的請求
if (isOneway) {
this.mQClientFactory.getMQClientAPIImpl().updateConsumerOffsetOneway(
findBrokerResult.getBrokerAddr(), requestHeader, 1000 * 5);
} else {
this.mQClientFactory.getMQClientAPIImpl().updateConsumerOffset(
findBrokerResult.getBrokerAddr(), requestHeader, 1000 * 5);
}
} else {
throw new MQClientException("The broker[" + mq.getBrokerName() + "] not exist", null);
}
}
}
主從模式下的消費進度同步
BrokerController
在建構函式中,例項化了SlaveSynchronize
,並在start方法中呼叫了handleSlaveSynchronize
方法處理從節點的資料同步,
如果當前的Broker是從節點,會註冊定時任務,定時呼叫SlaveSynchronize
的syncAll方法進行資料同步:
public class BrokerController {
private final SlaveSynchronize slaveSynchronize;
public BrokerController(
final BrokerConfig brokerConfig,
final NettyServerConfig nettyServerConfig,
final NettyClientConfig nettyClientConfig,
final MessageStoreConfig messageStoreConfig
) {
// ...
this.slaveSynchronize = new SlaveSynchronize(this);
//...
}
public void start() throws Exception {
if (!messageStoreConfig.isEnableDLegerCommitLog()) {
startProcessorByHa(messageStoreConfig.getBrokerRole());
// 處理從節點的同步
handleSlaveSynchronize(messageStoreConfig.getBrokerRole());
this.registerBrokerAll(true, false, true);
}
}
private void handleSlaveSynchronize(BrokerRole role) {
// 如果是SLAVE節點
if (role == BrokerRole.SLAVE) {
if (null != slaveSyncFuture) {
slaveSyncFuture.cancel(false);
}
this.slaveSynchronize.setMasterAddr(null);
// 設定定時任務,定時進行資料同步
slaveSyncFuture = this.scheduledExecutorService.scheduleAtFixedRate(new Runnable() {
@Override
public void run() {
try {
// 同步資料
BrokerController.this.slaveSynchronize.syncAll();
}
catch (Throwable e) {
log.error("ScheduledTask SlaveSynchronize syncAll error.", e);
}
}
}, 1000 * 3, 1000 * 10, TimeUnit.MILLISECONDS);
} else {
//handle the slave synchronise
if (null != slaveSyncFuture) {
slaveSyncFuture.cancel(false);
}
this.slaveSynchronize.setMasterAddr(null);
}
}
}
在SlaveSynchronize的syncAll
方法中,又呼叫了syncConsumerOffset
方法同步消費進度:
- 向主節點傳送請求獲取消費進度資料;
- 從節點將獲取到的消費進度資料進行持久化;
public class SlaveSynchronize {
public void syncAll() {
this.syncTopicConfig();
// 同步消費進度
this.syncConsumerOffset();
this.syncDelayOffset();
this.syncSubscriptionGroupConfig();
}
private void syncConsumerOffset() {
String masterAddrBak = this.masterAddr;
if (masterAddrBak != null && !masterAddrBak.equals(brokerController.getBrokerAddr())) {
try {
// 向主節點傳送請求獲取消費進度資訊
ConsumerOffsetSerializeWrapper offsetWrapper =
this.brokerController.getBrokerOuterAPI().getAllConsumerOffset(masterAddrBak);
// 設定資料
this.brokerController.getConsumerOffsetManager().getOffsetTable()
.putAll(offsetWrapper.getOffsetTable());
// 將獲取到的消費進度資料進行持久化
this.brokerController.getConsumerOffsetManager().persist();
log.info("Update slave consumer offset from master, {}", masterAddrBak);
} catch (Exception e) {
log.error("SyncConsumerOffset Exception, {}", masterAddrBak, e);
}
}
}
}
參考
丁威、周繼鋒《RocketMQ技術內幕》
RocketMQ 主從同步若干問題答疑
RocketMq 訂閱分組建立和刪除
RocketMQ版本:4.9.3