Dolphinscheduler Master模块源码剖析

1、Master Slot计算

核心代码逻辑org.apache.dolphinscheduler.server.master.registry.MasterSlotManager.SlotChangeListener#notify

public void notify(Map<String, MasterHeartBeat> masterNodeInfo) {
    List<Server> serverList = masterNodeInfo.values().stream()
            // TODO 这里其实就是过滤掉buzy的master节点
            .filter(heartBeat -> !heartBeat.getServerStatus().equals(ServerStatus.BUSY))
            .map(this::convertHeartBeatToServer).collect(Collectors.toList());
    // TODO 同步master节点
    syncMasterNodes(serverList);
}

计算 totalSlot和currentSlot

private void syncMasterNodes(List<Server> masterNodes) {
    slotLock.lock();
    try {
        this.masterPriorityQueue.clear();
        // TODO 这里会把所有的master节点都放入到masterPriorityQueue中，比如说 [192.168.220.1:12345,192.168.220.2:12345]
        this.masterPriorityQueue.putAll(masterNodes);
        // TODO 就是获取到本地ip的在队列中的位置
        int tempCurrentSlot = masterPriorityQueue.getIndex(masterConfig.getMasterAddress());
        // TODO 所有节点数量
        int tempTotalSlot = masterNodes.size();
        // TODO 正常情况下不会小于0
        if (tempCurrentSlot < 0) {
            totalSlot = 0;
            currentSlot = 0;
            log.warn("Current master is not in active master list");
        } else if (tempCurrentSlot != currentSlot || tempTotalSlot != totalSlot) {
            // TODO 这里其实就是记录的是比如说一共有两个slot，我的slot是0或者1
            totalSlot = tempTotalSlot;
            currentSlot = tempCurrentSlot;
            log.info("Update master nodes, total master size: {}, current slot: {}", totalSlot, currentSlot);
        }
    } finally {
        slotLock.unlock();
    }
}

this.masterPriorityQueue.putAll(masterNodes); 会计算索引

public void putAll(Collection<Server> serverList) {
    for (Server server : serverList) {
        this.queue.put(server);
    }
    // TODO 这里更新了hostIndexMap，存放的是 <host:port> -> 索引
    refreshMasterList();
}


private void refreshMasterList() {
    hostIndexMap.clear();
    Iterator<Server> iterator = queue.iterator();
    int index = 0;
    while (iterator.hasNext()) {
        Server server = iterator.next();
        String addr = NetUtils.getAddr(server.getHost(), server.getPort());
        hostIndexMap.put(addr, index);
        index += 1;
    }

}

2、Master消费Command生成流程实例

command最终的获取逻辑 :

比如说两个Master节点 : 
masterCount=2 thisMasterSlot=0  master1
masterCount=2 thisMasterSlot=1  master2

command中的数据如下 :
1 master2
2 master1
3 master2
4 master1

select *
        from t_ds_command
        where id % #{masterCount} = #{thisMasterSlot}
        order by process_instance_priority, id asc
            limit #{limit}

有没有感到疑惑，就是如果一个master更新到的最新的，一个没有更新到，怎么办？

比如说，master1节点是这样的
1  master2
2  master1
3  master2
4  master1

比如说，master2节点是这样的，是不是发现master2节点都是他的，都可以拉取消费？那就导致重复消费，比如说1这个command
1 master1
2 master1
3 master1
4 master1

org.apache.dolphinscheduler.service.process.ProcessServiceImpl#handleCommand

@Transactional
public @Nullable ProcessInstance handleCommand(String host,
                                                   Command command) throws CronParseException, CodeGenerateException {
    // TODO 创建流程实例
    ProcessInstance processInstance = constructProcessInstance(command, host);
    // cannot construct process instance, return null
    if (processInstance == null) {
        log.error("scan command, command parameter is error: {}", command);
        commandService.moveToErrorCommand(command, "process instance is null");
        return null;
    }
    processInstance.setCommandType(command.getCommandType());
    processInstance.addHistoryCmd(command.getCommandType());
    processInstance.setTestFlag(command.getTestFlag());
    // if the processDefinition is serial
    ProcessDefinition processDefinition = this.findProcessDefinition(processInstance.getProcessDefinitionCode(),
            processInstance.getProcessDefinitionVersion());
    // TODO 是否是串行执行
    if (processDefinition.getExecutionType().typeIsSerial()) {
        saveSerialProcess(processInstance, processDefinition);
        if (processInstance.getState() != WorkflowExecutionStatus.RUNNING_EXECUTION) {
            setSubProcessParam(processInstance);
            triggerRelationService.saveProcessInstanceTrigger(command.getId(), processInstance.getId());
            deleteCommandWithCheck(command.getId());
            // todo: this is a bad design to return null here, whether trigger the task
            return null;
        }
    } else {
        // TODO 并行执行
        processInstanceDao.upsertProcessInstance(processInstance);
    }

    // TODO 这里其实还会向triggerRelation表中插入一条数据，是流程实例和triggerCode的关系
    triggerRelationService.saveProcessInstanceTrigger(command.getId(), processInstance.getId());
    // TODO 设置子流程参数
    setSubProcessParam(processInstance);
    // TODO 删除command
    deleteCommandWithCheck(command.getId());
    return processInstance;
}

注意 : 这个方法是加@Transactional的，所以说创建流程实例和删除Command是在一个事物里面的，如果不同的Master消费到同一个Command。肯定会有一个删除Command失败，这时会抛出一个异常，这样就会让数据库进行回滚

3、工作流启动流程

4、DAG切分 & 任务提交

5、Master事件状态流转

图连接 : Master事件状态流转

TaskEventService组件中的TaskEventDispatchThread(线程)和TaskEventHandlerThread(线程)解析

其实就是Master自己状态(DISPATCH)和Worker汇报上来的状态(RUNNING、UPDATE_PID、RESULT)都会放入到eventQueue，TaskEventDispatchThread(线程)会阻塞的方式进行获取，然后放入到对应的TaskExecuteRunnable中(注意 : 不执行的)，只有通过TaskEventHandlerThread(线程)才会使用TaskExecuteThreadPool线程进行TaskExecuteRunnable的提交

ThreadPoolTaskExecutor的submitListenable参考
Spring ThreadPoolTaskExecutor解说

如感兴趣，点赞加关注，谢谢!!!