Oozie4.3
一 简介
1 官网
http://oozie.apache.org/
Apache Oozie Workflow Scheduler for Hadoop
Hadoop生态的工作流调度器
Overview
Oozie is a workflow scheduler system to manage Apache Hadoop jobs.
Oozie Workflow jobs are Directed Acyclical Graphs (DAGs) of actions.
Oozie Coordinator jobs are recurrent Oozie Workflow jobs triggered by time (frequency) and data availability.
Oozie is integrated with the rest of the Hadoop stack supporting several types of Hadoop jobs out of the box (such as Java map-reduce, Streaming map-reduce, Pig, Hive, Sqoop and Distcp) as well as system specific jobs (such as Java programs and shell scripts).
Oozie is a scalable, reliable and extensible system.
2 部署
3 数据库表结构
wf_jobs：工作流实例
wf_actions：任务实例
coord_jobs：调度实例
coord_actions：调度任务实例
4 概念
l Control Node：工作流的开始、结束以及决定Workflow的执行路径的节点（start、end、kill、decision、fork/join）
l Action Node：工作流执行的计算任务，支持的类型包括（HDFS、MapReduce、Java、Shell、SSH、Pig、Hive、E-Mail、Sub-Workflow、Sqoop、Distcp），即任务
l Workflow：由Control Node以及一系列Action Node组成的工作流，即工作流
l Coordinator：根据指定Cron信息触发workflow，即调度
l Bundle：按照组的方式批量管理Coordinator任务，实现集中的启停
二 代码解析
1 启动过程
加载配置的所有service：
ServicesLoader.contextInitialized
Services.init
Services.loadServices (oozie.services, oozie.services.ext)
Service结构：
Service
org.apache.oozie.service.SchedulerService,
org.apache.oozie.service.InstrumentationService,
org.apache.oozie.service.MemoryLocksService,
org.apache.oozie.service.UUIDService,
org.apache.oozie.service.ELService,
org.apache.oozie.service.AuthorizationService,
org.apache.oozie.service.UserGroupInformationService,
org.apache.oozie.service.HadoopAccessorService,
org.apache.oozie.service.JobsConcurrencyService,
org.apache.oozie.service.URIHandlerService,
org.apache.oozie.service.DagXLogInfoService,
org.apache.oozie.service.SchemaService,
org.apache.oozie.service.LiteWorkflowAppService,
org.apache.oozie.service.JPAService,
org.apache.oozie.service.StoreService,
org.apache.oozie.service.SLAStoreService,
org.apache.oozie.service.DBLiteWorkflowStoreService,
org.apache.oozie.service.CallbackService,
org.apache.oozie.service.ActionService,
org.apache.oozie.service.ShareLibService,
org.apache.oozie.service.CallableQueueService,
org.apache.oozie.service.ActionCheckerService,
org.apache.oozie.service.RecoveryService,
org.apache.oozie.service.PurgeService,
org.apache.oozie.service.CoordinatorEngineService,
org.apache.oozie.service.BundleEngineService,
org.apache.oozie.service.DagEngineService,
org.apache.oozie.service.CoordMaterializeTriggerService,
org.apache.oozie.service.StatusTransitService,
org.apache.oozie.service.PauseTransitService,
org.apache.oozie.service.GroupsService,
org.apache.oozie.service.ProxyUserService,
org.apache.oozie.service.XLogStreamingService,
org.apache.oozie.service.JvmPauseMonitorService,
org.apache.oozie.service.SparkConfigurationService
2 核心引擎
BaseEngine
DAGEngine （负责workflow执行）
CoordinatorEngine （负责coordinator执行）
BundleEngine （负责bundle执行）
3 workflow提交执行过程
DAGEngine.submitJob| submitJobFromCoordinator （提交workflow）
SubmitXCommand.call
execute
LiteWorkflowAppService.parseDef （解析得到WorkflowApp）
LiteWorkflowLib.parseDef
LiteWorkflowAppParser.validateAndParse
parse
WorkflowLib.createInstance （创建WorkflowInstance）
BatchQueryExecutor.executeBatchInsertUpdateDelete （保存WorkflowJobBean 到wf_jobs）
StartXCommand.call
SignalXCommand.call
execute
WorkflowInstance.start
LiteWorkflowInstance.start
signal
NodeHandler.enter
ActionNodeHandler.enter
start
LiteWorkflowStoreService.liteExecute (添加WorkflowActionBean到ACTIONS_TO_START)
WorkflowStoreService.getActionsToStart （从ACTIONS_TO_START取Action）
ActionStartXCommand.call
ActionExecutor.start
WorkflowNotificationXCommand.call
BatchQueryExecutor.executeBatchInsertUpdateDelete （保存WorkflowActionBean到wf_actions）
ActionExecutor.start是异步的，还需要检查Action执行状态来推进流程，有两种情况：
一种是Oozie Server正常运行：利用JobEndNotification
CallbackServlet.doGet
DagEngine.processCallback
CompletedActionXCommand.call
ActionCheckXCommand.call
ActionExecutor.check
ActionEndXCommand.call
SignalXCommand.call
一种是Oozie Server重启：利用ActionCheckerService
ActionCheckerService.init
ActionCheckRunnable.run
runWFActionCheck (GET_RUNNING_ACTIONS, oozie.service.ActionCheckerService.action.check.delay=600)
ActionCheckXCommand.call
ActionExecutor.check
ActionEndXCommand.call
SignalXCommand.call
runCoordActionCheck
4 coordinator提交执行过程
CoordinatorEngine.submitJob（提交coordinator）
CoordSubmitXCommand.call
submit
submitJob
storeToDB
CoordJobQueryExecutor.insert (保存CoordinatorJobBean到coord_jobs)
queueMaterializeTransitionXCommand
CoordMaterializeTransitionXCommand.call
execute
materialize
materializeActions
CoordCommandUtils.materializeOneInstance（创建CoordinatorActionBean）
storeToDB
performWrites
BatchQueryExecutor.executeBatchInsertUpdateDelete（保存CoordinatorActionBean到coord_actions）
CoordActionInputCheckXCommand.call
CoordActionReadyXCommand.call
CoordActionStartXCommand.call
DAGEngine.submitJobFromCoordinator
定时任务触发Materialize：
CoordMaterializeTriggerService.init
CoordMaterializeTriggerRunnable.run
CoordMaterializeTriggerService.runCoordJobMatLookup
materializeCoordJobs (GET_COORD_JOBS_OLDER_FOR_MATERIALIZATION)
CoordMaterializeTransitionXCommand.call
5 分布式
有些内部任务只能启动一个，单server环境Oozie中通过MemoryLocksService来保证，多server环境Oozie通过ZKLocksService来保证，要开启ZK，需要开启一些service：
org.apache.oozie.service.ZKLocksService,
org.apache.oozie.service.ZKXLogStreamingService,
org.apache.oozie.service.ZKJobsConcurrencyService,
org.apache.oozie.service.ZKUUIDService
同时需要配置oozie.zookeeper.connection.string
6 任务执行过程
ActionExecutor是任务执行的核心抽象基类，所有的具体任务都是这个类的子类
ActionExecutor
JavaActionExecutor
SshActionExecutor
FsActionExecutor
SubWorkflowActionExecutor
其中JavaActionExecutor是最重要的一个子类，很多其他的任务都是这个类的子类（比如HiveActionExecutor、SparkActionExecutor等）
JavaActionExecutor.start
prepareActionDir
submitLauncher
JobClient.getJob
injectLauncherCallback
ActionExecutor.Context.getCallbackUrl
job.end.notification.url
createLauncherConf
LauncherMapperHelper.setupLauncherInfo
JobClient.submitJob
check
JavaActionExecutor执行时会提交一个map任务到yarn，即LauncherMapper，
LauncherMapper.map
LauncherMain.main
LauncherMain是具体任务的执行类
LauncherMain
JavaMain
HiveMain
Hive2Main
SparkMain
ShellMain
SqoopMain