ZooKeeper（五）：事务处理之更新数据逻辑解析
通过前些文章，我们已经完全从整体架构和数据接入方面理解了ZK的前情工作。接下来，我们就来看ZK的正式工作吧。
本文以 setData /a data 这个命令作为出发点，来观察zk是如何处理来自客户端的数据更新操作的吧！
首先，我们需要明确各个角色所担任的工作，然后才能更好的理解其工作流程。我们以 Leader 为当前客户端连接的角色，揭开其 更新数据的面纱。其责任链是这样的：
LeaderRequestProcessor -> PrepRequestProcessor -> ProposalRequestProcessor -> CommitProcessor -> ToBeAppliedRequestProcessor -> FinalRequestProcessor
当然了， LeaderRequestProcessor 仅起到一个流程转发的作用，我们也不管它了。所以，起点是 PrepRequestProcessor 。
一、 PrepRequestProcessor 初步处理数据
有一点明确的是，每个独立运行的 RequestProcessor, 都是通过队列与一个 RequestProcessor 通信的。 PrepRequestProcessor 也一样，在接到 LeaderRequestProcessor sumit 过来的数据后，开始自己的工作。
// 提交任务为直接添加到队列过程 // org.apache.zookeeper.server.PrepRequestProcessor#processRequest public void processRequest(Request request) { request.prepQueueStartTime = Time.currentElapsedTime(); submittedRequests.add(request); ServerMetrics.getMetrics().PREP_PROCESSOR_QUEUED.add(1); } // org.apache.zookeeper.server.PrepRequestProcessor#run @Override public void run() { try { // 永远是死循环处理任务 while (true) { ServerMetrics.getMetrics().PREP_PROCESSOR_QUEUE_SIZE.add(submittedRequests.size()); // submittedRequests 阻塞队列获取 Request request = submittedRequests.take(); ServerMetrics.getMetrics().PREP_PROCESSOR_QUEUE_TIME .add(Time.currentElapsedTime() - request.prepQueueStartTime); long traceMask = ZooTrace.CLIENT_REQUEST_TRACE_MASK; if (request.type == OpCode.ping) { traceMask = ZooTrace.CLIENT_PING_TRACE_MASK; } if (LOG.isTraceEnabled()) { ZooTrace.logRequest(LOG, traceMask, 'P', request, ""); } if (Request.requestOfDeath == request) { break; } request.prepStartTime = Time.currentElapsedTime(); // 主要处理逻辑交由 pRequest pRequest(request); } } catch (Exception e) { handleException(this.getName(), e); } LOG.info("PrepRequestProcessor exited loop!"); } /** * This method will be called inside the ProcessRequestThread, which is a * singleton, so there will be a single thread calling this code. * * @param request */ protected void pRequest(Request request) throws RequestProcessorException { // LOG.info("Prep>>> cxid = " + request.cxid + " type = " + // request.type + " id = 0x" + Long.toHexString(request.sessionId)); request.setHdr(null); request.setTxn(null); try { switch (request.type) { case OpCode.createContainer: case OpCode.create: case OpCode.create2: CreateRequest create2Request = new CreateRequest(); pRequest2Txn(request.type, zks.getNextZxid(), request, create2Request, true); break; case OpCode.createTTL: CreateTTLRequest createTtlRequest = new CreateTTLRequest(); pRequest2Txn(request.type, zks.getNextZxid(), request, createTtlRequest, true); break; case OpCode.deleteContainer: case OpCode.delete: DeleteRequest deleteRequest = new DeleteRequest(); pRequest2Txn(request.type, zks.getNextZxid(), request, deleteRequest, true); break; // setData 的 type = 5 case OpCode.setData: SetDataRequest setDataRequest = new SetDataRequest(); // 针写数据请求，转交给 2Txn 处理 pRequest2Txn(request.type, zks.getNextZxid(), request, setDataRequest, true); break; case OpCode.reconfig: ReconfigRequest reconfigRequest = new ReconfigRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, reconfigRequest); pRequest2Txn(request.type, zks.getNextZxid(), request, reconfigRequest, true); break; case OpCode.setACL: SetACLRequest setAclRequest = new SetACLRequest(); pRequest2Txn(request.type, zks.getNextZxid(), request, setAclRequest, true); break; case OpCode.check: CheckVersionRequest checkRequest = new CheckVersionRequest(); pRequest2Txn(request.type, zks.getNextZxid(), request, checkRequest, true); break; case OpCode.multi: MultiOperationRecord multiRequest = new MultiOperationRecord(); try { ByteBufferInputStream.byteBuffer2Record(request.request, multiRequest); } catch (IOException e) { request.setHdr(new TxnHeader(request.sessionId, request.cxid, zks.getNextZxid(), Time.currentWallTime(), OpCode.multi)); throw e; } List<Txn> txns = new ArrayList<Txn>(); //Each op in a multi-op must have the same zxid! long zxid = zks.getNextZxid(); KeeperException ke = null; //Store off current pending change records in case we need to rollback Map<String, ChangeRecord> pendingChanges = getPendingChanges(multiRequest); for (Op op : multiRequest) { Record subrequest = op.toRequestRecord(); int type; Record txn; /* If we've already failed one of the ops, don't bother * trying the rest as we know it's going to fail and it * would be confusing in the logfiles. */ if (ke != null) { type = OpCode.error; txn = new ErrorTxn(Code.RUNTIMEINCONSISTENCY.intValue()); } else { /* Prep the request and convert to a Txn */ try { pRequest2Txn(op.getType(), zxid, request, subrequest, false); type = request.getHdr().getType(); txn = request.getTxn(); } catch (KeeperException e) { ke = e; type = OpCode.error; txn = new ErrorTxn(e.code().intValue()); if (e.code().intValue() > Code.APIERROR.intValue()) { LOG.info("Got user-level KeeperException when processing {} aborting" + " remaining multi ops. Error Path:{} Error:{}", request.toString(), e.getPath(), e.getMessage()); } request.setException(e); /* Rollback change records from failed multi-op */ rollbackPendingChanges(zxid, pendingChanges); } } // TODO: I don't want to have to serialize it here and then // immediately deserialize in next processor. But I'm // not sure how else to get the txn stored into our list. try (ByteArrayOutputStream baos = new ByteArrayOutputStream()) { BinaryOutputArchive boa = BinaryOutputArchive.getArchive(baos); txn.serialize(boa, "request"); ByteBuffer bb = ByteBuffer.wrap(baos.toByteArray()); txns.add(new Txn(type, bb.array())); } } request.setHdr(new TxnHeader(request.sessionId, request.cxid, zxid, Time.currentWallTime(), request.type)); request.setTxn(new MultiTxn(txns)); break; //create/close session don't require request record case OpCode.createSession: case OpCode.closeSession: if (!request.isLocalSession()) { pRequest2Txn(request.type, zks.getNextZxid(), request, null, true); } break; //All the rest don't need to create a Txn - just verify session case OpCode.sync: case OpCode.exists: case OpCode.getData: case OpCode.getACL: case OpCode.getChildren: case OpCode.getAllChildrenNumber: case OpCode.getChildren2: case OpCode.ping: case OpCode.setWatches: case OpCode.checkWatches: case OpCode.removeWatches: case OpCode.getEphemerals: case OpCode.multiRead: zks.sessionTracker.checkSession(request.sessionId, request.getOwner()); break; default: LOG.warn("unknown type {}", request.type); break; } } catch (KeeperException e) { if (request.getHdr() != null) { request.getHdr().setType(OpCode.error); request.setTxn(new ErrorTxn(e.code().intValue())); } if (e.code().intValue() > Code.APIERROR.intValue()) { LOG.info( "Got user-level KeeperException when processing {} Error Path:{} Error:{}", request.toString(), e.getPath(), e.getMessage()); } request.setException(e); } catch (Exception e) { // log at error level as we are returning a marshalling // error to the user LOG.error("Failed to process {}", request, e); StringBuilder sb = new StringBuilder(); ByteBuffer bb = request.request; if (bb != null) { bb.rewind(); while (bb.hasRemaining()) { sb.append(Integer.toHexString(bb.get() & 0xff)); } } else { sb.append("request buffer is null"); } LOG.error("Dumping request buffer: 0x{}", sb.toString()); if (request.getHdr() != null) { request.getHdr().setType(OpCode.error); request.setTxn(new ErrorTxn(Code.MARSHALLINGERROR.intValue())); } } // 将新的事务id获取到，传入下一个 RequestProcessor request.zxid = zks.getZxid(); ServerMetrics.getMetrics().PREP_PROCESS_TIME.add(Time.currentElapsedTime() - request.prepStartTime); // 这里的 nextProcessor 是 ProposalRequestProcessor nextProcessor.processRequest(request); } /** * This method will be called inside the ProcessRequestThread, which is a * singleton, so there will be a single thread calling this code. * * @param type * @param zxid * @param request * @param record */ protected void pRequest2Txn(int type, long zxid, Request request, Record record, boolean deserialize) throws KeeperException, IOException, RequestProcessorException { request.setHdr(new TxnHeader(request.sessionId, request.cxid, zxid, Time.currentWallTime(), type)); // 再次进行类型判定，确认处理逻辑 switch (type) { case OpCode.create: case OpCode.create2: case OpCode.createTTL: case OpCode.createContainer: { pRequest2TxnCreate(type, request, record, deserialize); break; } case OpCode.deleteContainer: { // ... break; } case OpCode.delete: //... break; // 同样我们只看 setData 情形 case OpCode.setData: // 分几步走: // 1. 检查session; // 2. 反序列化数据; // 3. 路径检查; // 4. 检查权限; // 5. 检查版本号; // 6. 添加到变更队列中，备用 zks.sessionTracker.checkSession(request.sessionId, request.getOwner()); SetDataRequest setDataRequest = (SetDataRequest) record; if (deserialize) { ByteBufferInputStream.byteBuffer2Record(request.request, setDataRequest); } path = setDataRequest.getPath(); validatePath(path, request.sessionId); nodeRecord = getRecordForPath(path); zks.checkACL(request.cnxn, nodeRecord.acl, ZooDefs.Perms.WRITE, request.authInfo, path, null); int newVersion = checkAndIncVersion(nodeRecord.stat.getVersion(), setDataRequest.getVersion(), path); request.setTxn(new SetDataTxn(path, setDataRequest.getData(), newVersion)); nodeRecord = nodeRecord.duplicate(request.getHdr().getZxid()); nodeRecord.stat.setVersion(newVersion); addChangeRecord(nodeRecord); break; case OpCode.reconfig: //... addChangeRecord(nodeRecord); break; case OpCode.setACL: // ... addChangeRecord(nodeRecord); break; case OpCode.createSession: // ... break; case OpCode.closeSession: // ... break; case OpCode.check: // ... break; default: LOG.warn("unknown type {}", type); break; } }
二、 ProposalRequestProcessor 处理写事务数据，投票发起
ProposalRequestProcessor 不作为独立的线程运行，只是在处理写事务时被调用。其入口也只有 processRequest 方法而已，主要作用则是组件投票逻辑。
// org.apache.zookeeper.server.quorum.ProposalRequestProcessor#processRequest public void processRequest(Request request) throws RequestProcessorException { // LOG.warn("Ack>>> cxid = " + request.cxid + " type = " + // request.type + " id = " + request.sessionId); // request.addRQRec(">prop"); /* In the following IF-THEN-ELSE block, we process syncs on the leader. * If the sync is coming from a follower, then the follower * handler adds it to syncHandler. Otherwise, if it is a client of * the leader that issued the sync command, then syncHandler won't * contain the handler. In this case, we add it to syncHandler, and * call processRequest on the next processor. */ if (request instanceof LearnerSyncRequest) { zks.getLeader().processSync((LearnerSyncRequest) request); } else { // 此处 nextProcessor 是 CommitProcessor // 先交其处理，然后再判断是否需要本地化数据 nextProcessor.processRequest(request); // 只要是写事务， hdr 不会为空 if (request.getHdr() != null) { // We need to sync and get consensus on any transactions // 投票决定是否更新数据 try { // 投票逻辑则由 leader 类去处理 zks.getLeader().propose(request); } catch (XidRolloverException e) { throw new RequestProcessorException(e.getMessage(), e); } // 交由 SyncRequestProcessor 进行数据持久化操作 syncProcessor.processRequest(request); } } } // 提交给 CommitProcessor 的任务队列 // org.apache.zookeeper.server.quorum.CommitProcessor#processRequest @Override public void processRequest(Request request) { if (stopped) { return; } LOG.debug("Processing request:: {}", request); request.commitProcQueueStartTime = Time.currentElapsedTime(); queuedRequests.add(request); // If the request will block, add it to the queue of blocking requests if (needCommit(request)) { queuedWriteRequests.add(request); numWriteQueuedRequests.incrementAndGet(); } else { numReadQueuedRequests.incrementAndGet(); } wakeup(); } // 简单看一下事务投票过程 // org.apache.zookeeper.server.quorum.Leader#propose /** * create a proposal and send it out to all the members * * @param request * @return the proposal that is queued to send to all the members */ public Proposal propose(Request request) throws XidRolloverException { /** * Address the rollover issue. All lower 32bits set indicate a new leader * election. Force a re-election instead. See ZOOKEEPER-1277 */ if ((request.zxid & 0xffffffffL) == 0xffffffffL) { String msg = "zxid lower 32 bits have rolled over, forcing re-election, and therefore new epoch start"; shutdown(msg); throw new XidRolloverException(msg); } // 封装 QuorumPacket, 进行节点间通信 byte[] data = SerializeUtils.serializeRequest(request); proposalStats.setLastBufferSize(data.length); QuorumPacket pp = new QuorumPacket(Leader.PROPOSAL, request.zxid, data, null); Proposal p = new Proposal(); p.packet = pp; p.request = request; synchronized (this) { // 将自己算入投票队列 p.addQuorumVerifier(self.getQuorumVerifier()); if (request.getHdr().getType() == OpCode.reconfig) { self.setLastSeenQuorumVerifier(request.qv, true); } if (self.getQuorumVerifier().getVersion() < self.getLastSeenQuorumVerifier().getVersion()) { p.addQuorumVerifier(self.getLastSeenQuorumVerifier()); } LOG.debug("Proposing:: {}", request); lastProposed = p.packet.getZxid(); outstandingProposals.put(lastProposed, p); // 此处会将 QuorumPacket 数据包，全部发送到各个通信节点 sendPacket(pp); } ServerMetrics.getMetrics().PROPOSAL_COUNT.add(1); return p; } // 同步处理器的加入逻辑，也是将其放入队列中，然后自行轮询队列处理 // org.apache.zookeeper.server.SyncRequestProcessor#processRequest public void processRequest(final Request request) { Objects.requireNonNull(request, "Request cannot be null"); request.syncQueueStartTime = Time.currentElapsedTime(); queuedRequests.add(request); ServerMetrics.getMetrics().SYNC_PROCESSOR_QUEUED.add(1); }
三、 CommitProcessor 处理数据的提交操作，触发ack等
由 PrepRequestProcessor 提交任务过来，自身独立处理request.
// org.apache.zookeeper.server.quorum.CommitProcessor#processRequest @Override public void processRequest(Request request) { if (stopped) { return; } LOG.debug("Processing request:: {}", request); request.commitProcQueueStartTime = Time.currentElapsedTime(); queuedRequests.add(request); // If the request will block, add it to the queue of blocking requests if (needCommit(request)) { queuedWriteRequests.add(request); numWriteQueuedRequests.incrementAndGet(); } else { numReadQueuedRequests.incrementAndGet(); } wakeup(); } // org.apache.zookeeper.server.quorum.CommitProcessor#run @Override public void run() { try { /* * In each iteration of the following loop we process at most * requestsToProcess requests of queuedRequests. We have to limit * the number of request we poll from queuedRequests, since it is * possible to endlessly poll read requests from queuedRequests, and * that will lead to a starvation of non-local committed requests. */ int requestsToProcess = 0; boolean commitIsWaiting = false; do { /* * Since requests are placed in the queue before being sent to * the leader, if commitIsWaiting = true, the commit belongs to * the first update operation in the queuedRequests or to a * request from a client on another server (i.e., the order of * the following two lines is important!). */ commitIsWaiting = !committedRequests.isEmpty(); requestsToProcess = queuedRequests.size(); // Avoid sync if we have something to do // wait/notify 等待任务处理完成 if (requestsToProcess == 0 && !commitIsWaiting) { // Waiting for requests to process synchronized (this) { while (!stopped && requestsToProcess == 0 && !commitIsWaiting) { wait(); commitIsWaiting = !committedRequests.isEmpty(); requestsToProcess = queuedRequests.size(); } } } ServerMetrics.getMetrics().READS_QUEUED_IN_COMMIT_PROCESSOR.add(numReadQueuedRequests.get()); ServerMetrics.getMetrics().WRITES_QUEUED_IN_COMMIT_PROCESSOR.add(numWriteQueuedRequests.get()); ServerMetrics.getMetrics().COMMITS_QUEUED_IN_COMMIT_PROCESSOR.add(committedRequests.size()); long time = Time.currentElapsedTime(); /* * Processing up to requestsToProcess requests from the incoming * queue (queuedRequests). If maxReadBatchSize is set then no * commits will be processed until maxReadBatchSize number of * reads are processed (or no more reads remain in the queue). * After the loop a single committed request is processed if * one is waiting (or a batch of commits if maxCommitBatchSize * is set). */ Request request; int readsProcessed = 0; // 处理单批次数据，由前面取得的 requestsToProcess 决定 while (!stopped && requestsToProcess > 0 && (maxReadBatchSize < 0 || readsProcessed <= maxReadBatchSize) && (request = queuedRequests.poll()) != null) { // 将队列弹出后，就相当于处理了该任务 requestsToProcess--; // 更新类操作将会进入第一个逻辑 if (needCommit(request) || pendingRequests.containsKey(request.sessionId)) { // Add request to pending Deque<Request> requests = pendingRequests.computeIfAbsent(request.sessionId, sid -> new ArrayDeque<>()); requests.addLast(request); ServerMetrics.getMetrics().REQUESTS_IN_SESSION_QUEUE.add(requests.size()); } else { readsProcessed++; numReadQueuedRequests.decrementAndGet(); sendToNextProcessor(request); } /* * Stop feeding the pool if there is a local pending update * and a committed request that is ready. Once we have a * pending request with a waiting committed request, we know * we can process the committed one. This is because commits * for local requests arrive in the order they appeared in * the queue, so if we have a pending request and a * committed request, the committed request must be for that * pending write or for a write originating at a different * server. We skip this if maxReadBatchSize is set. */ if (maxReadBatchSize < 0 && !pendingRequests.isEmpty() && !committedRequests.isEmpty()) { /* * We set commitIsWaiting so that we won't check * committedRequests again. */ commitIsWaiting = true; break; } } ServerMetrics.getMetrics().READS_ISSUED_IN_COMMIT_PROC.add(readsProcessed); if (!commitIsWaiting) { commitIsWaiting = !committedRequests.isEmpty(); } /* * Handle commits, if any. */ if (commitIsWaiting && !stopped) { /* * Drain outstanding reads */ waitForEmptyPool(); if (stopped) { return; } int commitsToProcess = maxCommitBatchSize; /* * Loop through all the commits, and try to drain them. */ Set<Long> queuesToDrain = new HashSet<>(); long startWriteTime = Time.currentElapsedTime(); int commitsProcessed = 0; while (commitIsWaiting && !stopped && commitsToProcess > 0) { // Process committed head request = committedRequests.peek(); /* * Check if this is a local write request is pending, * if so, update it with the committed info. If the commit matches * the first write queued in the blockedRequestQueue, we know this is * a commit for a local write, as commits are received in order. Else * it must be a commit for a remote write. */ if (!queuedWriteRequests.isEmpty() && queuedWriteRequests.peek().sessionId == request.sessionId && queuedWriteRequests.peek().cxid == request.cxid) { /* * Commit matches the earliest write in our write queue. */ Deque<Request> sessionQueue = pendingRequests.get(request.sessionId); ServerMetrics.getMetrics().PENDING_SESSION_QUEUE_SIZE.add(pendingRequests.size()); if (sessionQueue == null || sessionQueue.isEmpty() || !needCommit(sessionQueue.peek())) { /* * Can't process this write yet. * Either there are reads pending in this session, or we * haven't gotten to this write yet. */ break; } else { ServerMetrics.getMetrics().REQUESTS_IN_SESSION_QUEUE.add(sessionQueue.size()); // If session queue != null, then it is also not empty. Request topPending = sessionQueue.poll(); /* * Generally, we want to send to the next processor our version of the request, * since it contains the session information that is needed for post update processing. * In more details, when a request is in the local queue, there is (or could be) a client * attached to this server waiting for a response, and there is other bookkeeping of * requests that are outstanding and have originated from this server * (e.g., for setting the max outstanding requests) - we need to update this info when an * outstanding request completes. Note that in the other case, the operation * originated from a different server and there is no local bookkeeping or a local client * session that needs to be notified. */ topPending.setHdr(request.getHdr()); topPending.setTxn(request.getTxn()); topPending.zxid = request.zxid; topPending.commitRecvTime = request.commitRecvTime; request = topPending; // Only decrement if we take a request off the queue. numWriteQueuedRequests.decrementAndGet(); queuedWriteRequests.poll(); queuesToDrain.add(request.sessionId); } } /* * Pull the request off the commit queue, now that we are going * to process it. */ committedRequests.remove(); commitsToProcess--; commitsProcessed++; // Process the write inline. // 处理 commit, ToBeAppliedRequestProcessor processWrite(request); commitIsWaiting = !committedRequests.isEmpty(); } ServerMetrics.getMetrics().WRITE_BATCH_TIME_IN_COMMIT_PROCESSOR .add(Time.currentElapsedTime() - startWriteTime); ServerMetrics.getMetrics().WRITES_ISSUED_IN_COMMIT_PROC.add(commitsProcessed); /* * Process following reads if any, remove session queue(s) if * empty. */ readsProcessed = 0; for (Long sessionId : queuesToDrain) { Deque<Request> sessionQueue = pendingRequests.get(sessionId); int readsAfterWrite = 0; while (!stopped && !sessionQueue.isEmpty() && !needCommit(sessionQueue.peek())) { numReadQueuedRequests.decrementAndGet(); // 构造 org.apache.zookeeper.server.quorum.CommitProcessor.CommitWorkRequest#CommitWorkRequest, 提交到线程池运行 sendToNextProcessor(sessionQueue.poll()); readsAfterWrite++; } ServerMetrics.getMetrics().READS_AFTER_WRITE_IN_SESSION_QUEUE.add(readsAfterWrite); readsProcessed += readsAfterWrite; // Remove empty queues if (sessionQueue.isEmpty()) { pendingRequests.remove(sessionId); } } ServerMetrics.getMetrics().SESSION_QUEUES_DRAINED.add(queuesToDrain.size()); ServerMetrics.getMetrics().READ_ISSUED_FROM_SESSION_QUEUE.add(readsProcessed); } ServerMetrics.getMetrics().COMMIT_PROCESS_TIME.add(Time.currentElapsedTime() - time); endOfIteration(); } while (!stoppedMainLoop); } catch (Throwable e) { handleException(this.getName(), e); } LOG.info("CommitProcessor exited loop!"); } // 当commitProcessor 检测到可以进行事务提交时，就会进行提交动作（这个动作是在投票已结束的时候才开始的） // org.apache.zookeeper.server.quorum.CommitProcessor#processWrite private void processWrite(Request request) throws RequestProcessorException { processCommitMetrics(request, true); long timeBeforeFinalProc = Time.currentElapsedTime(); // ToBeAppliedRequestProcessor nextProcessor.processRequest(request); ServerMetrics.getMetrics().WRITE_FINAL_PROC_TIME.add(Time.currentElapsedTime() - timeBeforeFinalProc); } // org.apache.zookeeper.server.quorum.Leader.ToBeAppliedRequestProcessor#processRequest public void processRequest(Request request) throws RequestProcessorException { // FinalRequestProcessor, 由 FinalRequestProcessor 进行数据持久化 next.processRequest(request); // The only requests that should be on toBeApplied are write // requests, for which we will have a hdr. We can't simply use // request.zxid here because that is set on read requests to equal // the zxid of the last write op. if (request.getHdr() != null) { long zxid = request.getHdr().getZxid(); Iterator<Proposal> iter = leader.toBeApplied.iterator(); if (iter.hasNext()) { Proposal p = iter.next(); if (p.request != null && p.request.zxid == zxid) { iter.remove(); return; } } LOG.error("Committed request not found on toBeApplied: {}", request); } } // org.apache.zookeeper.server.FinalRequestProcessor#processRequest public void processRequest(Request request) { LOG.debug("Processing request:: {}", request); // request.addRQRec(">final"); long traceMask = ZooTrace.CLIENT_REQUEST_TRACE_MASK; if (request.type == OpCode.ping) { traceMask = ZooTrace.SERVER_PING_TRACE_MASK; } if (LOG.isTraceEnabled()) { ZooTrace.logRequest(LOG, traceMask, 'E', request, ""); } // DB 入库 ProcessTxnResult rc = zks.processTxn(request); // ZOOKEEPER-558: // In some cases the server does not close the connection (e.g., closeconn buffer // was not being queued — ZOOKEEPER-558) properly. This happens, for example, // when the client closes the connection. The server should still close the session, though. // Calling closeSession() after losing the cnxn, results in the client close session response being dropped. if (request.type == OpCode.closeSession && connClosedByClient(request)) { // We need to check if we can close the session id. // Sometimes the corresponding ServerCnxnFactory could be null because // we are just playing diffs from the leader. if (closeSession(zks.serverCnxnFactory, request.sessionId) || closeSession(zks.secureServerCnxnFactory, request.sessionId)) { return; } } if (request.getHdr() != null) { /* * Request header is created only by the leader, so this must be * a quorum request. Since we're comparing timestamps across hosts, * this metric may be incorrect. However, it's still a very useful * metric to track in the happy case. If there is clock drift, * the latency can go negative. Note: headers use wall time, not * CLOCK_MONOTONIC. */ long propagationLatency = Time.currentWallTime() - request.getHdr().getTime(); if (propagationLatency >= 0) { ServerMetrics.getMetrics().PROPAGATION_LATENCY.add(propagationLatency); } } if (request.cnxn == null) { return; } ServerCnxn cnxn = request.cnxn; long lastZxid = zks.getZKDatabase().getDataTreeLastProcessedZxid(); String lastOp = "NA"; // Notify ZooKeeperServer that the request has finished so that it can // update any request accounting/throttling limits zks.decInProcess(); zks.requestFinished(request); Code err = Code.OK; Record rsp = null; String path = null; try { if (request.getHdr() != null && request.getHdr().getType() == OpCode.error) { /* * When local session upgrading is disabled, leader will * reject the ephemeral node creation due to session expire. * However, if this is the follower that issue the request, * it will have the correct error code, so we should use that * and report to user */ if (request.getException() != null) { throw request.getException(); } else { throw KeeperException.create(KeeperException.Code.get(((ErrorTxn) request.getTxn()).getErr())); } } KeeperException ke = request.getException(); if (ke instanceof SessionMovedException) { throw ke; } if (ke != null && request.type != OpCode.multi) { throw ke; } LOG.debug("{}", request); if (request.isStale()) { ServerMetrics.getMetrics().STALE_REPLIES.add(1); } switch (request.type) { case OpCode.ping: { lastOp = "PING"; updateStats(request, lastOp, lastZxid); cnxn.sendResponse(new ReplyHeader(-2, lastZxid, 0), null, "response"); return; } case OpCode.createSession: { lastOp = "SESS"; updateStats(request, lastOp, lastZxid); zks.finishSessionInit(request.cnxn, true); return; } case OpCode.multi: { lastOp = "MULT"; rsp = new MultiResponse(); for (ProcessTxnResult subTxnResult : rc.multiResult) { OpResult subResult; switch (subTxnResult.type) { case OpCode.check: subResult = new CheckResult(); break; case OpCode.create: subResult = new CreateResult(subTxnResult.path); break; case OpCode.create2: case OpCode.createTTL: case OpCode.createContainer: subResult = new CreateResult(subTxnResult.path, subTxnResult.stat); break; case OpCode.delete: case OpCode.deleteContainer: subResult = new DeleteResult(); break; case OpCode.setData: subResult = new SetDataResult(subTxnResult.stat); break; case OpCode.error: subResult = new ErrorResult(subTxnResult.err); if (subTxnResult.err == Code.SESSIONMOVED.intValue()) { throw new SessionMovedException(); } break; default: throw new IOException("Invalid type of op"); } ((MultiResponse) rsp).add(subResult); } break; } case OpCode.multiRead: { lastOp = "MLTR"; MultiOperationRecord multiReadRecord = new MultiOperationRecord(); ByteBufferInputStream.byteBuffer2Record(request.request, multiReadRecord); rsp = new MultiResponse(); OpResult subResult; for (Op readOp : multiReadRecord) { try { Record rec; switch (readOp.getType()) { case OpCode.getChildren: rec = handleGetChildrenRequest(readOp.toRequestRecord(), cnxn, request.authInfo); subResult = new GetChildrenResult(((GetChildrenResponse) rec).getChildren()); break; case OpCode.getData: rec = handleGetDataRequest(readOp.toRequestRecord(), cnxn, request.authInfo); GetDataResponse gdr = (GetDataResponse) rec; subResult = new GetDataResult(gdr.getData(), gdr.getStat()); break; default: throw new IOException("Invalid type of readOp"); } } catch (KeeperException e) { subResult = new ErrorResult(e.code().intValue()); } ((MultiResponse) rsp).add(subResult); } break; } case OpCode.create: { lastOp = "CREA"; rsp = new CreateResponse(rc.path); err = Code.get(rc.err); requestPathMetricsCollector.registerRequest(request.type, rc.path); break; } case OpCode.create2: case OpCode.createTTL: case OpCode.createContainer: { lastOp = "CREA"; rsp = new Create2Response(rc.path, rc.stat); err = Code.get(rc.err); requestPathMetricsCollector.registerRequest(request.type, rc.path); break; } case OpCode.delete: case OpCode.deleteContainer: { lastOp = "DELE"; err = Code.get(rc.err); requestPathMetricsCollector.registerRequest(request.type, rc.path); break; } case OpCode.setData: { // 写数据 lastOp = "SETD"; rsp = new SetDataResponse(rc.stat); err = Code.get(rc.err); requestPathMetricsCollector.registerRequest(request.type, rc.path); break; } case OpCode.reconfig: { lastOp = "RECO"; rsp = new GetDataResponse( ((QuorumZooKeeperServer) zks).self.getQuorumVerifier().toString().getBytes(), rc.stat); err = Code.get(rc.err); break; } case OpCode.setACL: { lastOp = "SETA"; rsp = new SetACLResponse(rc.stat); err = Code.get(rc.err); requestPathMetricsCollector.registerRequest(request.type, rc.path); break; } case OpCode.closeSession: { lastOp = "CLOS"; err = Code.get(rc.err); break; } case OpCode.sync: { lastOp = "SYNC"; SyncRequest syncRequest = new SyncRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, syncRequest); rsp = new SyncResponse(syncRequest.getPath()); requestPathMetricsCollector.registerRequest(request.type, syncRequest.getPath()); break; } case OpCode.check: { lastOp = "CHEC"; rsp = new SetDataResponse(rc.stat); err = Code.get(rc.err); break; } case OpCode.exists: { lastOp = "EXIS"; // TODO we need to figure out the security requirement for this! ExistsRequest existsRequest = new ExistsRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, existsRequest); path = existsRequest.getPath(); if (path.indexOf('\0') != -1) { throw new KeeperException.BadArgumentsException(); } Stat stat = zks.getZKDatabase().statNode(path, existsRequest.getWatch() ? cnxn : null); rsp = new ExistsResponse(stat); requestPathMetricsCollector.registerRequest(request.type, path); break; } case OpCode.getData: { lastOp = "GETD"; GetDataRequest getDataRequest = new GetDataRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, getDataRequest); path = getDataRequest.getPath(); rsp = handleGetDataRequest(getDataRequest, cnxn, request.authInfo); requestPathMetricsCollector.registerRequest(request.type, path); break; } case OpCode.setWatches: { lastOp = "SETW"; SetWatches setWatches = new SetWatches(); // TODO We really should NOT need this!!!! request.request.rewind(); ByteBufferInputStream.byteBuffer2Record(request.request, setWatches); long relativeZxid = setWatches.getRelativeZxid(); zks.getZKDatabase() .setWatches( relativeZxid, setWatches.getDataWatches(), setWatches.getExistWatches(), setWatches.getChildWatches(), cnxn); break; } case OpCode.getACL: { lastOp = "GETA"; GetACLRequest getACLRequest = new GetACLRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, getACLRequest); path = getACLRequest.getPath(); DataNode n = zks.getZKDatabase().getNode(path); if (n == null) { throw new KeeperException.NoNodeException(); } zks.checkACL( request.cnxn, zks.getZKDatabase().aclForNode(n), ZooDefs.Perms.READ | ZooDefs.Perms.ADMIN, request.authInfo, path, null); Stat stat = new Stat(); List<ACL> acl = zks.getZKDatabase().getACL(path, stat); requestPathMetricsCollector.registerRequest(request.type, getACLRequest.getPath()); try { zks.checkACL( request.cnxn, zks.getZKDatabase().aclForNode(n), ZooDefs.Perms.ADMIN, request.authInfo, path, null); rsp = new GetACLResponse(acl, stat); } catch (KeeperException.NoAuthException e) { List<ACL> acl1 = new ArrayList<ACL>(acl.size()); for (ACL a : acl) { if ("digest".equals(a.getId().getScheme())) { Id id = a.getId(); Id id1 = new Id(id.getScheme(), id.getId().replaceAll(":.*", ":x")); acl1.add(new ACL(a.getPerms(), id1)); } else { acl1.add(a); } } rsp = new GetACLResponse(acl1, stat); } break; } case OpCode.getChildren: { lastOp = "GETC"; GetChildrenRequest getChildrenRequest = new GetChildrenRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, getChildrenRequest); path = getChildrenRequest.getPath(); rsp = handleGetChildrenRequest(getChildrenRequest, cnxn, request.authInfo); requestPathMetricsCollector.registerRequest(request.type, path); break; } case OpCode.getAllChildrenNumber: { lastOp = "GETACN"; GetAllChildrenNumberRequest getAllChildrenNumberRequest = new GetAllChildrenNumberRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, getAllChildrenNumberRequest); path = getAllChildrenNumberRequest.getPath(); DataNode n = zks.getZKDatabase().getNode(path); if (n == null) { throw new KeeperException.NoNodeException(); } zks.checkACL( request.cnxn, zks.getZKDatabase().aclForNode(n), ZooDefs.Perms.READ, request.authInfo, path, null); int number = zks.getZKDatabase().getAllChildrenNumber(path); rsp = new GetAllChildrenNumberResponse(number); break; } case OpCode.getChildren2: { lastOp = "GETC"; GetChildren2Request getChildren2Request = new GetChildren2Request(); ByteBufferInputStream.byteBuffer2Record(request.request, getChildren2Request); Stat stat = new Stat(); path = getChildren2Request.getPath(); DataNode n = zks.getZKDatabase().getNode(path); if (n == null) { throw new KeeperException.NoNodeException(); } zks.checkACL( request.cnxn, zks.getZKDatabase().aclForNode(n), ZooDefs.Perms.READ, request.authInfo, path, null); List<String> children = zks.getZKDatabase() .getChildren(path, stat, getChildren2Request.getWatch() ? cnxn : null); rsp = new GetChildren2Response(children, stat); requestPathMetricsCollector.registerRequest(request.type, path); break; } case OpCode.checkWatches: { lastOp = "CHKW"; CheckWatchesRequest checkWatches = new CheckWatchesRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, checkWatches); WatcherType type = WatcherType.fromInt(checkWatches.getType()); path = checkWatches.getPath(); boolean containsWatcher = zks.getZKDatabase().containsWatcher(path, type, cnxn); if (!containsWatcher) { String msg = String.format(Locale.ENGLISH, "%s (type: %s)", path, type); throw new KeeperException.NoWatcherException(msg); } requestPathMetricsCollector.registerRequest(request.type, checkWatches.getPath()); break; } case OpCode.removeWatches: { lastOp = "REMW"; RemoveWatchesRequest removeWatches = new RemoveWatchesRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, removeWatches); WatcherType type = WatcherType.fromInt(removeWatches.getType()); path = removeWatches.getPath(); boolean removed = zks.getZKDatabase().removeWatch(path, type, cnxn); if (!removed) { String msg = String.format(Locale.ENGLISH, "%s (type: %s)", path, type); throw new KeeperException.NoWatcherException(msg); } requestPathMetricsCollector.registerRequest(request.type, removeWatches.getPath()); break; } case OpCode.getEphemerals: { lastOp = "GETE"; GetEphemeralsRequest getEphemerals = new GetEphemeralsRequest(); ByteBufferInputStream.byteBuffer2Record(request.request, getEphemerals); String prefixPath = getEphemerals.getPrefixPath(); Set<String> allEphems = zks.getZKDatabase().getDataTree().getEphemerals(request.sessionId); List<String> ephemerals = new ArrayList<>(); if (StringUtils.isBlank(prefixPath) || "/".equals(prefixPath.trim())) { ephemerals.addAll(allEphems); } else { for (String p : allEphems) { if (p.startsWith(prefixPath)) { ephemerals.add(p); } } } rsp = new GetEphemeralsResponse(ephemerals); break; } } } catch (SessionMovedException e) { // session moved is a connection level error, we need to tear // down the connection otw ZOOKEEPER-710 might happen // ie client on slow follower starts to renew session, fails // before this completes, then tries the fast follower (leader) // and is successful, however the initial renew is then // successfully fwd/processed by the leader and as a result // the client and leader disagree on where the client is most // recently attached (and therefore invalid SESSION MOVED generated) cnxn.sendCloseSession(); return; } catch (KeeperException e) { err = e.code(); } catch (Exception e) { // log at error level as we are returning a marshalling // error to the user LOG.error("Failed to process {}", request, e); StringBuilder sb = new StringBuilder(); ByteBuffer bb = request.request; bb.rewind(); while (bb.hasRemaining()) { sb.append(Integer.toHexString(bb.get() & 0xff)); } LOG.error("Dumping request buffer: 0x{}", sb.toString()); err = Code.MARSHALLINGERROR; } ReplyHeader hdr = new ReplyHeader(request.cxid, lastZxid, err.intValue()); // 更新状态 updateStats(request, lastOp, lastZxid); try { if (request.type == OpCode.getData && path != null && rsp != null) { // Serialized read responses could be cached by the connection object. // Cache entries are identified by their path and last modified zxid, // so these values are passed along with the response. GetDataResponse getDataResponse = (GetDataResponse) rsp; Stat stat = null; if (getDataResponse.getStat() != null) { stat = getDataResponse.getStat(); } cnxn.sendResponse(hdr, rsp, "response", path, stat); } else { // 返回数据给客户端 cnxn.sendResponse(hdr, rsp, "response"); } if (request.type == OpCode.closeSession) { cnxn.sendCloseSession(); } } catch (IOException e) { LOG.error("FIXMSG", e); } } // org.apache.zookeeper.server.ZooKeeperServer#processTxn // entry point for FinalRequestProcessor.java public ProcessTxnResult processTxn(Request request) { TxnHeader hdr = request.getHdr(); processTxnForSessionEvents(request, hdr, request.getTxn()); final boolean writeRequest = (hdr != null); final boolean quorumRequest = request.isQuorum(); // return fast w/o synchronization when we get a read if (!writeRequest && !quorumRequest) { return new ProcessTxnResult(); } synchronized (outstandingChanges) { ProcessTxnResult rc = processTxnInDB(hdr, request.getTxn()); // request.hdr is set for write requests, which are the only ones // that add to outstandingChanges. if (writeRequest) { long zxid = hdr.getZxid(); while (!outstandingChanges.isEmpty() && outstandingChanges.peek().zxid <= zxid) { ChangeRecord cr = outstandingChanges.remove(); ServerMetrics.getMetrics().OUTSTANDING_CHANGES_REMOVED.add(1); if (cr.zxid < zxid) { LOG.warn( "Zxid outstanding 0x{} is less than current 0x{}", Long.toHexString(cr.zxid), Long.toHexString(zxid)); } if (outstandingChangesForPath.get(cr.path) == cr) { outstandingChangesForPath.remove(cr.path); } } } // do not add non quorum packets to the queue. // 此处的队列只是为了方便 follower 能够快速复制数据 if (quorumRequest) { getZKDatabase().addCommittedProposal(request); } return rc; } }
四、Leader#propose 进行事务操作投票保证数据的高可用性
// 半数支持原则,主要是 组装 QuorumPacket, 然后发送给各个投票节点 // org.apache.zookeeper.server.quorum.Leader#propose /** * create a proposal and send it out to all the members * * @param request * @return the proposal that is queued to send to all the members */ public Proposal propose(Request request) throws XidRolloverException { /** * Address the rollover issue. All lower 32bits set indicate a new leader * election. Force a re-election instead. See ZOOKEEPER-1277 */ if ((request.zxid & 0xffffffffL) == 0xffffffffL) { String msg = "zxid lower 32 bits have rolled over, forcing re-election, and therefore new epoch start"; shutdown(msg); throw new XidRolloverException(msg); } byte[] data = SerializeUtils.serializeRequest(request); proposalStats.setLastBufferSize(data.length); QuorumPacket pp = new QuorumPacket(Leader.PROPOSAL, request.zxid, data, null); Proposal p = new Proposal(); p.packet = pp; p.request = request; synchronized (this) { p.addQuorumVerifier(self.getQuorumVerifier()); if (request.getHdr().getType() == OpCode.reconfig) { self.setLastSeenQuorumVerifier(request.qv, true); } if (self.getQuorumVerifier().getVersion() < self.getLastSeenQuorumVerifier().getVersion()) { p.addQuorumVerifier(self.getLastSeenQuorumVerifier()); } LOG.debug("Proposing:: {}", request); lastProposed = p.packet.getZxid(); outstandingProposals.put(lastProposed, p); // 发送数据包给各投票节点 sendPacket(pp); } ServerMetrics.getMetrics().PROPOSAL_COUNT.add(1); return p; } /** * send a packet to all the followers ready to follow * * @param qp * the packet to be sent */ void sendPacket(QuorumPacket qp) { synchronized (forwardingFollowers) { for (LearnerHandler f : forwardingFollowers) { // 添加到 LearnerHandler 的队列，就返回，可见 LearnerHandler 肯定又是一个异步任务 f.queuePacket(qp); } } } // org.apache.zookeeper.server.quorum.LearnerHandler#queuePacket void queuePacket(QuorumPacket p) { queuedPackets.add(p); // Add a MarkerQuorumPacket at regular intervals. // 周期性的放一个 Marker 到队列中 if (shouldSendMarkerPacketForLogging() && packetCounter.getAndIncrement() % markerPacketInterval == 0) { queuedPackets.add(new MarkerQuorumPacket(System.nanoTime())); } queuedPacketsSize.addAndGet(packetSize(p)); }
五、 LearnerHandler 投票通信线程处理提交过来的投票队列
上一节我们看到，Leader 只是交事务投票的包添加到 LearnerHandler 的队列后就返回了，所以， LearnerHandler 肯定有自己另外的处理逻辑。否则如何跟其他节点通信呢？
// org.apache.zookeeper.server.quorum.LearnerHandler#sendPackets /** * This method will use the thread to send packets added to the * queuedPackets list * * @throws InterruptedException */ private void sendPackets() throws InterruptedException { long traceMask = ZooTrace.SERVER_PACKET_TRACE_MASK; // 一个死循环一直处理发送队列 while (true) { try { QuorumPacket p; p = queuedPackets.poll(); if (p == null) { bufferedOutput.flush(); p = queuedPackets.take(); } ServerMetrics.getMetrics().LEARNER_HANDLER_QP_SIZE.add(Long.toString(this.sid), queuedPackets.size()); // 前面看到会定期的提交 MarkerQuorumPacket 过来，但这里会忽略掉，无实际影响 if (p instanceof MarkerQuorumPacket) { MarkerQuorumPacket m = (MarkerQuorumPacket) p; ServerMetrics.getMetrics().LEARNER_HANDLER_QP_TIME .add(Long.toString(this.sid), (System.nanoTime() - m.time) / 1000000L); continue; } queuedPacketsSize.addAndGet(-packetSize(p)); if (p == proposalOfDeath) { // Packet of death! break; } // ping 包 if (p.getType() == Leader.PING) { traceMask = ZooTrace.SERVER_PING_TRACE_MASK; } // 投票包，我们要观察的包在这里 if (p.getType() == Leader.PROPOSAL) { syncLimitCheck.updateProposal(p.getZxid(), System.nanoTime()); } if (LOG.isTraceEnabled()) { ZooTrace.logQuorumPacket(LOG, traceMask, 'o', p); } // Log the zxid of the last request, if it is a valid zxid. if (p.getZxid() > 0) { lastZxid = p.getZxid(); } // 往socket中写入投票请求 oa.writeRecord(p, "packet"); packetsSent.incrementAndGet(); messageTracker.trackSent(p.getType()); } catch (IOException e) { if (!sock.isClosed()) { LOG.warn("Unexpected exception at {}", this, e); try { // this will cause everything to shutdown on // this learner handler and will help notify // the learner/observer instantaneously sock.close(); } catch (IOException ie) { LOG.warn("Error closing socket for handler {}", this, ie); } } break; } } } // 上面这个发送线程，是在 LearnerHandler 被调用时生成的 /** * Start thread that will forward any packet in the queue to the follower */ protected void startSendingPackets() { if (!sendingThreadStarted) { // Start sending packets new Thread() { public void run() { Thread.currentThread().setName("Sender-" + sock.getRemoteSocketAddress()); try { sendPackets(); } catch (InterruptedException e) { LOG.warn("Unexpected interruption", e); } } }.start(); sendingThreadStarted = true; } else { LOG.error("Attempting to start sending thread after it already started"); } } /** * This thread will receive packets from the peer and process them and * also listen to new connections from new peers. */ @Override public void run() { try { learnerMaster.addLearnerHandler(this); tickOfNextAckDeadline = learnerMaster.getTickOfInitialAckDeadline(); ia = BinaryInputArchive.getArchive(bufferedInput); bufferedOutput = new BufferedOutputStream(sock.getOutputStream()); oa = BinaryOutputArchive.getArchive(bufferedOutput); QuorumPacket qp = new QuorumPacket(); ia.readRecord(qp, "packet"); messageTracker.trackReceived(qp.getType()); if (qp.getType() != Leader.FOLLOWERINFO && qp.getType() != Leader.OBSERVERINFO) { LOG.error("First packet {} is not FOLLOWERINFO or OBSERVERINFO!", qp.toString()); return; } if (learnerMaster instanceof ObserverMaster && qp.getType() != Leader.OBSERVERINFO) { throw new IOException("Non observer attempting to connect to ObserverMaster. type = " + qp.getType()); } byte[] learnerInfoData = qp.getData(); if (learnerInfoData != null) { ByteBuffer bbsid = ByteBuffer.wrap(learnerInfoData); if (learnerInfoData.length >= 8) { this.sid = bbsid.getLong(); } if (learnerInfoData.length >= 12) { this.version = bbsid.getInt(); // protocolVersion } if (learnerInfoData.length >= 20) { long configVersion = bbsid.getLong(); if (configVersion > learnerMaster.getQuorumVerifierVersion()) { throw new IOException("Follower is ahead of the leader (has a later activated configuration)"); } } } else { this.sid = learnerMaster.getAndDecrementFollowerCounter(); } String followerInfo = learnerMaster.getPeerInfo(this.sid); if (followerInfo.isEmpty()) { LOG.info( "Follower sid: {} not in the current config {}", this.sid, Long.toHexString(learnerMaster.getQuorumVerifierVersion())); } else { LOG.info("Follower sid: {} : info : {}", this.sid, followerInfo); } if (qp.getType() == Leader.OBSERVERINFO) { learnerType = LearnerType.OBSERVER; } learnerMaster.registerLearnerHandlerBean(this, sock); long lastAcceptedEpoch = ZxidUtils.getEpochFromZxid(qp.getZxid()); long peerLastZxid; StateSummary ss = null; long zxid = qp.getZxid(); long newEpoch = learnerMaster.getEpochToPropose(this.getSid(), lastAcceptedEpoch); long newLeaderZxid = ZxidUtils.makeZxid(newEpoch, 0); if (this.getVersion() < 0x10000) { // we are going to have to extrapolate the epoch information long epoch = ZxidUtils.getEpochFromZxid(zxid); ss = new StateSummary(epoch, zxid); // fake the message learnerMaster.waitForEpochAck(this.getSid(), ss); } else { byte[] ver = new byte[4]; ByteBuffer.wrap(ver).putInt(0x10000); QuorumPacket newEpochPacket = new QuorumPacket(Leader.LEADERINFO, newLeaderZxid, ver, null); oa.writeRecord(newEpochPacket, "packet"); messageTracker.trackSent(Leader.LEADERINFO); bufferedOutput.flush(); QuorumPacket ackEpochPacket = new QuorumPacket(); ia.readRecord(ackEpochPacket, "packet"); messageTracker.trackReceived(ackEpochPacket.getType()); if (ackEpochPacket.getType() != Leader.ACKEPOCH) { LOG.error("{} is not ACKEPOCH", ackEpochPacket.toString()); return; } ByteBuffer bbepoch = ByteBuffer.wrap(ackEpochPacket.getData()); ss = new StateSummary(bbepoch.getInt(), ackEpochPacket.getZxid()); learnerMaster.waitForEpochAck(this.getSid(), ss); } peerLastZxid = ss.getLastZxid(); // Take any necessary action if we need to send TRUNC or DIFF // startForwarding() will be called in all cases boolean needSnap = syncFollower(peerLastZxid, learnerMaster); // syncs between followers and the leader are exempt from throttling because it // is importatnt to keep the state of quorum servers up-to-date. The exempted syncs // are counted as concurrent syncs though boolean exemptFromThrottle = getLearnerType() != LearnerType.OBSERVER; /* if we are not truncating or sending a diff just send a snapshot */ if (needSnap) { syncThrottler = learnerMaster.getLearnerSnapSyncThrottler(); syncThrottler.beginSync(exemptFromThrottle); try { long zxidToSend = learnerMaster.getZKDatabase().getDataTreeLastProcessedZxid(); oa.writeRecord(new QuorumPacket(Leader.SNAP, zxidToSend, null, null), "packet"); messageTracker.trackSent(Leader.SNAP); bufferedOutput.flush(); LOG.info( "Sending snapshot last zxid of peer is 0x{}, zxid of leader is 0x{}, " + "send zxid of db as 0x{}, {} concurrent snapshot sync, " + "snapshot sync was {} from throttle", Long.toHexString(peerLastZxid), Long.toHexString(leaderLastZxid), Long.toHexString(zxidToSend), syncThrottler.getSyncInProgress(), exemptFromThrottle ? "exempt" : "not exempt"); // Dump data to peer learnerMaster.getZKDatabase().serializeSnapshot(oa); oa.writeString("BenWasHere", "signature"); bufferedOutput.flush(); } finally { ServerMetrics.getMetrics().SNAP_COUNT.add(1); } } else { syncThrottler = learnerMaster.getLearnerDiffSyncThrottler(); syncThrottler.beginSync(exemptFromThrottle); ServerMetrics.getMetrics().DIFF_COUNT.add(1); } LOG.debug("Sending NEWLEADER message to {}", sid); // the version of this quorumVerifier will be set by leader.lead() in case // the leader is just being established. waitForEpochAck makes sure that readyToStart is true if // we got here, so the version was set if (getVersion() < 0x10000) { QuorumPacket newLeaderQP = new QuorumPacket(Leader.NEWLEADER, newLeaderZxid, null, null); oa.writeRecord(newLeaderQP, "packet"); } else { QuorumPacket newLeaderQP = new QuorumPacket(Leader.NEWLEADER, newLeaderZxid, learnerMaster.getQuorumVerifierBytes(), null); queuedPackets.add(newLeaderQP); } bufferedOutput.flush(); // Start thread that blast packets in the queue to learner startSendingPackets(); /* * Have to wait for the first ACK, wait until * the learnerMaster is ready, and only then we can * start processing messages. */ qp = new QuorumPacket(); ia.readRecord(qp, "packet"); messageTracker.trackReceived(qp.getType()); if (qp.getType() != Leader.ACK) { LOG.error("Next packet was supposed to be an ACK, but received packet: {}", packetToString(qp)); return; } LOG.debug("Received NEWLEADER-ACK message from {}", sid); learnerMaster.waitForNewLeaderAck(getSid(), qp.getZxid()); syncLimitCheck.start(); // sync ends when NEWLEADER-ACK is received syncThrottler.endSync(); syncThrottler = null; // now that the ack has been processed expect the syncLimit sock.setSoTimeout(learnerMaster.syncTimeout()); /* * Wait until learnerMaster starts up */ learnerMaster.waitForStartup(); // Mutation packets will be queued during the serialize, // so we need to mark when the peer can actually start // using the data // LOG.debug("Sending UPTODATE message to {}", sid); queuedPackets.add(new QuorumPacket(Leader.UPTODATE, -1, null, null)); while (true) { qp = new QuorumPacket(); ia.readRecord(qp, "packet"); messageTracker.trackReceived(qp.getType()); long traceMask = ZooTrace.SERVER_PACKET_TRACE_MASK; if (qp.getType() == Leader.PING) { traceMask = ZooTrace.SERVER_PING_TRACE_MASK; } if (LOG.isTraceEnabled()) { ZooTrace.logQuorumPacket(LOG, traceMask, 'i', qp); } tickOfNextAckDeadline = learnerMaster.getTickOfNextAckDeadline(); packetsReceived.incrementAndGet(); ByteBuffer bb; long sessionId; int cxid; int type; switch (qp.getType()) { case Leader.ACK: if (this.learnerType == LearnerType.OBSERVER) { LOG.debug("Received ACK from Observer {}", this.sid); } syncLimitCheck.updateAck(qp.getZxid()); // 提交 ack, 即收到了一提交的请求，则调用 Leader 进行处理 learnerMaster.processAck(this.sid, qp.getZxid(), sock.getLocalSocketAddress()); break; case Leader.PING: // Process the touches ByteArrayInputStream bis = new ByteArrayInputStream(qp.getData()); DataInputStream dis = new DataInputStream(bis); while (dis.available() > 0) { long sess = dis.readLong(); int to = dis.readInt(); learnerMaster.touch(sess, to); } break; case Leader.REVALIDATE: ServerMetrics.getMetrics().REVALIDATE_COUNT.add(1); learnerMaster.revalidateSession(qp, this); break; case Leader.REQUEST: bb = ByteBuffer.wrap(qp.getData()); sessionId = bb.getLong(); cxid = bb.getInt(); type = bb.getInt(); bb = bb.slice(); Request si; if (type == OpCode.sync) { si = new LearnerSyncRequest(this, sessionId, cxid, type, bb, qp.getAuthinfo()); } else { si = new Request(null, sessionId, cxid, type, bb, qp.getAuthinfo()); } si.setOwner(this); learnerMaster.submitLearnerRequest(si); requestsReceived.incrementAndGet(); break; default: LOG.warn("unexpected quorum packet, type: {}", packetToString(qp)); break; } } } catch (IOException e) { if (sock != null && !sock.isClosed()) { LOG.error("Unexpected exception causing shutdown while sock still open", e); //close the socket to make sure the //other side can see it being close try { sock.close(); } catch (IOException ie) { // do nothing } } } catch (InterruptedException e) { LOG.error("Unexpected exception in LearnerHandler.", e); } catch (SyncThrottleException e) { LOG.error("too many concurrent sync.", e); syncThrottler = null; } catch (Exception e) { LOG.error("Unexpected exception in LearnerHandler.", e); throw e; } finally { if (syncThrottler != null) { syncThrottler.endSync(); syncThrottler = null; } String remoteAddr = getRemoteAddress(); LOG.warn("******* GOODBYE {} ********", remoteAddr); messageTracker.dumpToLog(remoteAddr); shutdown(); } } // org.apache.zookeeper.server.quorum.Leader#processAck /** * Keep a count of acks that are received by the leader for a particular * proposal * * @param zxid, the zxid of the proposal sent out * @param sid, the id of the server that sent the ack * @param followerAddr */ @Override public synchronized void processAck(long sid, long zxid, SocketAddress followerAddr) { if (!allowedToCommit) { return; // last op committed was a leader change - from now on } // the new leader should commit if (LOG.isTraceEnabled()) { LOG.trace("Ack zxid: 0x{}", Long.toHexString(zxid)); for (Proposal p : outstandingProposals.values()) { long packetZxid = p.packet.getZxid(); LOG.trace("outstanding proposal: 0x{}", Long.toHexString(packetZxid)); } LOG.trace("outstanding proposals all"); } if ((zxid & 0xffffffffL) == 0) { /* * We no longer process NEWLEADER ack with this method. However, * the learner sends an ack back to the leader after it gets * UPTODATE, so we just ignore the message. */ return; } if (outstandingProposals.size() == 0) { LOG.debug("outstanding is 0"); return; } if (lastCommitted >= zxid) { LOG.debug( "proposal has already been committed, pzxid: 0x{} zxid: 0x{}", Long.toHexString(lastCommitted), Long.toHexString(zxid)); // The proposal has already been committed return; } Proposal p = outstandingProposals.get(zxid); if (p == null) { LOG.warn("Trying to commit future proposal: zxid 0x{} from {}", Long.toHexString(zxid), followerAddr); return; } if (ackLoggingFrequency > 0 && (zxid % ackLoggingFrequency == 0)) { p.request.logLatency(ServerMetrics.getMetrics().ACK_LATENCY, Long.toString(sid)); } p.addAck(sid); // 尝试 commit boolean hasCommitted = tryToCommit(p, zxid, followerAddr); // If p is a reconfiguration, multiple other operations may be ready to be committed, // since operations wait for different sets of acks. // Currently we only permit one outstanding reconfiguration at a time // such that the reconfiguration and subsequent outstanding ops proposed while the reconfig is // pending all wait for a quorum of old and new config, so its not possible to get enough acks // for an operation without getting enough acks for preceding ops. But in the future if multiple // concurrent reconfigs are allowed, this can happen and then we need to check whether some pending // ops may already have enough acks and can be committed, which is what this code does. if (hasCommitted && p.request != null && p.request.getHdr().getType() == OpCode.reconfig) { long curZxid = zxid; while (allowedToCommit && hasCommitted && p != null) { curZxid++; p = outstandingProposals.get(curZxid); if (p != null) { hasCommitted = tryToCommit(p, curZxid, null); } } } } // org.apache.zookeeper.server.quorum.Leader#tryToCommit /** * @return True if committed, otherwise false. **/ public synchronized boolean tryToCommit(Proposal p, long zxid, SocketAddress followerAddr) { // make sure that ops are committed in order. With reconfigurations it is now possible // that different operations wait for different sets of acks, and we still want to enforce // that they are committed in order. Currently we only permit one outstanding reconfiguration // such that the reconfiguration and subsequent outstanding ops proposed while the reconfig is // pending all wait for a quorum of old and new config, so it's not possible to get enough acks // for an operation without getting enough acks for preceding ops. But in the future if multiple // concurrent reconfigs are allowed, this can happen. if (outstandingProposals.containsKey(zxid - 1)) { return false; } // in order to be committed, a proposal must be accepted by a quorum. // // getting a quorum from all necessary configurations. if (!p.hasAllQuorums()) { return false; } // commit proposals in order if (zxid != lastCommitted + 1) { LOG.warn( "Commiting zxid 0x{} from {} noy first!", Long.toHexString(zxid), followerAddr); LOG.warn("First is {}", (lastCommitted + 1)); } outstandingProposals.remove(zxid); if (p.request != null) { toBeApplied.add(p); } if (p.request == null) { LOG.warn("Going to commit null: {}", p); } else if (p.request.getHdr().getType() == OpCode.reconfig) { LOG.debug("Committing a reconfiguration! {}", outstandingProposals.size()); //if this server is voter in new config with the same quorum address, //then it will remain the leader //otherwise an up-to-date follower will be designated as leader. This saves //leader election time, unless the designated leader fails Long designatedLeader = getDesignatedLeader(p, zxid); //LOG.warn("designated leader is: " + designatedLeader); QuorumVerifier newQV = p.qvAcksetPairs.get(p.qvAcksetPairs.size() - 1).getQuorumVerifier(); self.processReconfig(newQV, designatedLeader, zk.getZxid(), true); if (designatedLeader != self.getId()) { allowedToCommit = false; } // we're sending the designated leader, and if the leader is changing the followers are // responsible for closing the connection - this way we are sure that at least a majority of them // receive the commit message. commitAndActivate(zxid, designatedLeader); informAndActivate(p, designatedLeader); //turnOffFollowers(); } else { p.request.logLatency(ServerMetrics.getMetrics().QUORUM_ACK_LATENCY); commit(zxid); inform(p); } // 依次提交事务 request, 其实就是放到 CommitProcessor 的提交队列 zk.commitProcessor.commit(p.request); if (pendingSyncs.containsKey(zxid)) { // 为 Learner 发送 SYNC 通知 for (LearnerSyncRequest r : pendingSyncs.remove(zxid)) { sendSync(r); } } return true; }
六、 SyncRequestProcessor 数据持久化处理器
// org.apache.zookeeper.server.quorum.ProposalRequestProcessor#processRequest public void processRequest(Request request) throws RequestProcessorException { // LOG.warn("Ack>>> cxid = " + request.cxid + " type = " + // request.type + " id = " + request.sessionId); // request.addRQRec(">prop"); /* In the following IF-THEN-ELSE block, we process syncs on the leader. * If the sync is coming from a follower, then the follower * handler adds it to syncHandler. Otherwise, if it is a client of * the leader that issued the sync command, then syncHandler won't * contain the handler. In this case, we add it to syncHandler, and * call processRequest on the next processor. */ if (request instanceof LearnerSyncRequest) { zks.getLeader().processSync((LearnerSyncRequest) request); } else { nextProcessor.processRequest(request); if (request.getHdr() != null) { // We need to sync and get consensus on any transactions try { zks.getLeader().propose(request); } catch (XidRolloverException e) { throw new RequestProcessorException(e.getMessage(), e); } // 只有写事务，才需要进行持久化操作 syncProcessor.processRequest(request); } } } // org.apache.zookeeper.server.SyncRequestProcessor#processRequest public void processRequest(final Request request) { Objects.requireNonNull(request, "Request cannot be null"); request.syncQueueStartTime = Time.currentElapsedTime(); // 添加到队列后返回 queuedRequests.add(request); ServerMetrics.getMetrics().SYNC_PROCESSOR_QUEUED.add(1); } // org.apache.zookeeper.server.SyncRequestProcessor#run @Override public void run() { try { // we do this in an attempt to ensure that not all of the servers // in the ensemble take a snapshot at the same time resetSnapshotStats(); lastFlushTime = Time.currentElapsedTime(); while (true) { ServerMetrics.getMetrics().SYNC_PROCESSOR_QUEUE_SIZE.add(queuedRequests.size()); long pollTime = Math.min(zks.getMaxWriteQueuePollTime(), getRemainingDelay()); // 从 queuedRequests 队列取数据 Request si = queuedRequests.poll(pollTime, TimeUnit.MILLISECONDS); if (si == null) { /* We timed out looking for more writes to batch, go ahead and flush immediately */ flush(); si = queuedRequests.take(); } if (si == REQUEST_OF_DEATH) { break; } long startProcessTime = Time.currentElapsedTime(); ServerMetrics.getMetrics().SYNC_PROCESSOR_QUEUE_TIME.add(startProcessTime - si.syncQueueStartTime); // track the number of records written to the log // 添加到 事务日志中 if (zks.getZKDatabase().append(si)) { if (shouldSnapshot()) { resetSnapshotStats(); // roll the log zks.getZKDatabase().rollLog(); // take a snapshot if (!snapThreadMutex.tryAcquire()) { LOG.warn("Too busy to snap, skipping"); } else { new ZooKeeperThread("Snapshot Thread") { public void run() { try { zks.takeSnapshot(); } catch (Exception e) { LOG.warn("Unexpected exception", e); } finally { snapThreadMutex.release(); } } }.start(); } } } else if (toFlush.isEmpty()) { // optimization for read heavy workloads // iff this is a read, and there are no pending // flushes (writes), then just pass this to the next // processor if (nextProcessor != null) { nextProcessor.processRequest(si); if (nextProcessor instanceof Flushable) { ((Flushable) nextProcessor).flush(); } } continue; } toFlush.add(si); // 2 // 刷写数据到磁盘，完成 Sync 操作 if (shouldFlush()) { flush(); } ServerMetrics.getMetrics().SYNC_PROCESS_TIME.add(Time.currentElapsedTime() - startProcessTime); } } catch (Throwable t) { handleException(this.getName(), t); } LOG.info("SyncRequestProcessor exited!"); } /** If both flushDelay and maxMaxBatchSize are set (bigger than 0), flush * whenever either condition is hit. If only one or the other is * set, flush only when the relevant condition is hit. */ private boolean shouldFlush() { long flushDelay = zks.getFlushDelay(); long maxBatchSize = zks.getMaxBatchSize(); if ((flushDelay > 0) && (getRemainingDelay() == 0)) { return true; } return (maxBatchSize > 0) && (toFlush.size() >= maxBatchSize); }
至此，一个数据的更新操作就完成了。看起来是有点复杂呢。
主要有:
1. 判断写事务;
2. 发起投票操作;
3. 发送 ack 回应;
4. 可以提交事务, 交给 CommitProcessor;
5. 刷写数据到磁盘;
七、 等等，各个事务的投票是如何处理的？
是的，前面我们看到通过队列，将请求包给到了 LearnerHandler 的两个线程。顶多也只看到了发送的过程，并未看到有什么超过半数投票之类的动作，它在哪里呢？
其实是在 tryToCommit 的时候进行判断的。
// org.apache.zookeeper.server.quorum.Leader#tryToCommit /** * @return True if committed, otherwise false. **/ public synchronized boolean tryToCommit(Proposal p, long zxid, SocketAddress followerAddr) { // make sure that ops are committed in order. With reconfigurations it is now possible // that different operations wait for different sets of acks, and we still want to enforce // that they are committed in order. Currently we only permit one outstanding reconfiguration // such that the reconfiguration and subsequent outstanding ops proposed while the reconfig is // pending all wait for a quorum of old and new config, so it's not possible to get enough acks // for an operation without getting enough acks for preceding ops. But in the future if multiple // concurrent reconfigs are allowed, this can happen. if (outstandingProposals.containsKey(zxid - 1)) { return false; } // in order to be committed, a proposal must be accepted by a quorum. // // getting a quorum from all necessary configurations. // 确认是否达到了可以通过的确认数量，如果没有则不提交。 // 如果超过了半数提交，则直接进入提交后续，当然，这会有并发控制 if (!p.hasAllQuorums()) { return false; } // commit proposals in order if (zxid != lastCommitted + 1) { LOG.warn( "Commiting zxid 0x{} from {} noy first!", Long.toHexString(zxid), followerAddr); LOG.warn("First is {}", (lastCommitted + 1)); } outstandingProposals.remove(zxid); if (p.request != null) { toBeApplied.add(p); } if (p.request == null) { LOG.warn("Going to commit null: {}", p); } else if (p.request.getHdr().getType() == OpCode.reconfig) { LOG.debug("Committing a reconfiguration! {}", outstandingProposals.size()); //if this server is voter in new config with the same quorum address, //then it will remain the leader //otherwise an up-to-date follower will be designated as leader. This saves //leader election time, unless the designated leader fails Long designatedLeader = getDesignatedLeader(p, zxid); //LOG.warn("designated leader is: " + designatedLeader); QuorumVerifier newQV = p.qvAcksetPairs.get(p.qvAcksetPairs.size() - 1).getQuorumVerifier(); self.processReconfig(newQV, designatedLeader, zk.getZxid(), true); if (designatedLeader != self.getId()) { allowedToCommit = false; } // we're sending the designated leader, and if the leader is changing the followers are // responsible for closing the connection - this way we are sure that at least a majority of them // receive the commit message. commitAndActivate(zxid, designatedLeader); informAndActivate(p, designatedLeader); //turnOffFollowers(); } else { p.request.logLatency(ServerMetrics.getMetrics().QUORUM_ACK_LATENCY); // 提交后，会把 lastCommitted 设置为当前 zxid // 会给 各节点发送 COMMIT 信息提交 commit(zxid); inform(p); } zk.commitProcessor.commit(p.request); if (pendingSyncs.containsKey(zxid)) { for (LearnerSyncRequest r : pendingSyncs.remove(zxid)) { sendSync(r); } } return true; } // org.apache.zookeeper.server.quorum.SyncedLearnerTracker#hasAllQuorums public boolean hasAllQuorums() { for (QuorumVerifierAcksetPair qvAckset : qvAcksetPairs) { // 此处是判断是否超过半数的请求被 ack if (!qvAckset.getQuorumVerifier().containsQuorum(qvAckset.getAckset())) { return false; } } // 如果超过半数 ack, 则进行提交动作 return true; } // org.apache.zookeeper.server.quorum.flexible.QuorumMaj#containsQuorum /** * Verifies if a set is a majority. Assumes that ackSet contains acks only * from votingMembers */ public boolean containsQuorum(Set<Long> ackSet) { return (ackSet.size() > half); } // 通知提交事务 // org.apache.zookeeper.server.quorum.Leader#commit /** * Create a commit packet and send it to all the members of the quorum * * @param zxid */ public void commit(long zxid) { synchronized (this) { lastCommitted = zxid; } QuorumPacket qp = new QuorumPacket(Leader.COMMIT, zxid, null, null); sendPacket(qp); ServerMetrics.getMetrics().COMMIT_COUNT.add(1); } // 从上面可以看出，每收到一个节点的响应，就会走一次提交判断逻辑，且只要超过一半就会进行提交，如果不作后续限制，则肯定会出重复提交问题，这个问题是在 processAck 里通过 zxid 来判断的 /** * Keep a count of acks that are received by the leader for a particular * proposal * * @param zxid, the zxid of the proposal sent out * @param sid, the id of the server that sent the ack * @param followerAddr */ @Override public synchronized void processAck(long sid, long zxid, SocketAddress followerAddr) { if (!allowedToCommit) { return; // last op committed was a leader change - from now on } // the new leader should commit if (LOG.isTraceEnabled()) { LOG.trace("Ack zxid: 0x{}", Long.toHexString(zxid)); for (Proposal p : outstandingProposals.values()) { long packetZxid = p.packet.getZxid(); LOG.trace("outstanding proposal: 0x{}", Long.toHexString(packetZxid)); } LOG.trace("outstanding proposals all"); } if ((zxid & 0xffffffffL) == 0) { /* * We no longer process NEWLEADER ack with this method. However, * the learner sends an ack back to the leader after it gets * UPTODATE, so we just ignore the message. */ return; } if (outstandingProposals.size() == 0) { LOG.debug("outstanding is 0"); return; } // 如果已提交，则 lastCommitted 会变大，从而使本次提交失效返回 if (lastCommitted >= zxid) { LOG.debug( "proposal has already been committed, pzxid: 0x{} zxid: 0x{}", Long.toHexString(lastCommitted), Long.toHexString(zxid)); // The proposal has already been committed return; } Proposal p = outstandingProposals.get(zxid); if (p == null) { LOG.warn("Trying to commit future proposal: zxid 0x{} from {}", Long.toHexString(zxid), followerAddr); return; } if (ackLoggingFrequency > 0 && (zxid % ackLoggingFrequency == 0)) { p.request.logLatency(ServerMetrics.getMetrics().ACK_LATENCY, Long.toString(sid)); } p.addAck(sid); boolean hasCommitted = tryToCommit(p, zxid, followerAddr); // If p is a reconfiguration, multiple other operations may be ready to be committed, // since operations wait for different sets of acks. // Currently we only permit one outstanding reconfiguration at a time // such that the reconfiguration and subsequent outstanding ops proposed while the reconfig is // pending all wait for a quorum of old and new config, so its not possible to get enough acks // for an operation without getting enough acks for preceding ops. But in the future if multiple // concurrent reconfigs are allowed, this can happen and then we need to check whether some pending // ops may already have enough acks and can be committed, which is what this code does. if (hasCommitted && p.request != null && p.request.getHdr().getType() == OpCode.reconfig) { long curZxid = zxid; while (allowedToCommit && hasCommitted && p != null) { curZxid++; p = outstandingProposals.get(curZxid); if (p != null) { hasCommitted = tryToCommit(p, curZxid, null); } } } }
至此，整体流程完成！Leader 任务完成， Follower 的流程略有差异。但是肯定没Leader的复杂，感兴趣同学可以去看看。
虽然一个看起来很简单的更新动作，在这里显得如此之复杂，但是在准确性和高可用面前，内部的复杂性是必要的。至少在你使用时只需set一下就可以了，却享受了api带给你的关键保障，这也是基础设施给我们带来的好处。
不要害怕今日的苦，你要相信明天，更苦！
