cn.ponfee.disjob.supervisor.component

类 DistributedJobManager

java.lang.Object

cn.ponfee.disjob.supervisor.component.AbstractJobManager

cn.ponfee.disjob.supervisor.component.DistributedJobManager

@Component
public class DistributedJobManager
extends AbstractJobManager

Manage distributed schedule job.

作者:

Ponfee

字段概要

从类继承的字段 cn.ponfee.disjob.supervisor.component.AbstractJobManager

conf, dependMapper, jobMapper, log

构造器概要

构造器

构造器和说明

DistributedJobManager(SupervisorProperties conf, SchedJobMapper jobMapper, SchedDependMapper dependMapper, SchedInstanceMapper instanceMapper, SchedTaskMapper taskMapper, SchedWorkflowMapper workflowMapper, cn.ponfee.disjob.common.base.IdGenerator idGenerator, SupervisorRegistry discoveryWorker, TaskDispatcher taskDispatcher, DiscoveryServerRestProxy.GroupedServerInvoker<cn.ponfee.disjob.core.base.WorkerRpcService> groupedWorkerRpcClient, DestinationServerRestProxy.DestinationServerInvoker<cn.ponfee.disjob.core.base.WorkerRpcService,cn.ponfee.disjob.core.base.Worker> destinationWorkerRpcClient, org.springframework.transaction.support.TransactionTemplate transactionTemplate)

方法概要

限定符和类型	方法和说明
protected void	abortBroadcastWaitingTask(long taskId) Broadcast strategy task after assigned worker.
boolean	cancelInstance(long instanceId, cn.ponfee.disjob.core.enums.Operation ops) Cancel instance
void	changeInstanceState(long instanceId, cn.ponfee.disjob.core.enums.ExecuteState toExecuteState) Force change instance state
boolean	createInstance(cn.ponfee.disjob.core.model.SchedJob job, TriggerInstance triggerInstance) Update sched job, save sched instance and tasks.
void	deleteInstance(long instanceId)
protected List<cn.ponfee.disjob.core.model.SchedTask>	listPausableTasks(long instanceId) Lists the pausable tasks
boolean	pauseInstance(long instanceId) Pause instance
void	processTaskDispatchFailedEvent(TaskDispatchFailedEvent event) Listen task dispatch failed event `@Async`需要标注`@EnableAsync`来启用，默认使用的是`SimpleAsyncTaskExecutor`线程池，会为每个任务创建一个新线程(慎用)
boolean	purgeInstance(cn.ponfee.disjob.core.model.SchedInstance inst) Purge the zombie instance which maybe dead
boolean	resumeInstance(long instanceId) Resume the instance from paused to waiting state
boolean	savepoint(long taskId, String worker, String executeSnapshot)
cn.ponfee.disjob.core.dto.supervisor.StartTaskResult	startTask(cn.ponfee.disjob.core.dto.supervisor.StartTaskParam param) Starts the task
boolean	stopTask(cn.ponfee.disjob.core.dto.supervisor.StopTaskParam param) Stop task
void	triggerJob(long jobId) Manual trigger the sched job
boolean	updateInstanceNextScanTime(cn.ponfee.disjob.core.model.SchedInstance instance, Date nextScanTime)
void	updateTaskWorker(String worker, List<Long> taskIds) Set or clear task worker 当worker不相同时，可使用`CASE WHEN`语法： UPDATE sched_task SET worker = CASE WHEN task_id = 1 THEN 'a' WHEN task_id = 2 THEN 'b' END WHERE id IN (1, 2)

从类继承的方法 cn.ponfee.disjob.supervisor.component.AbstractJobManager

addJob, changeJobState, deleteJob, disableJob, dispatch, dispatch, dispatch, generateId, hasAliveExecuting, hasNotDiscoveredWorkers, hasNotDiscoveredWorkers, isAliveWorker, isAliveWorker, isDeadWorker, shouldRedispatch, splitJob, updateFixedDelayNextTriggerTime, updateJob, updateJobNextScanTime, updateJobNextTriggerTime

从类继承的方法 java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

构造器详细资料

DistributedJobManager

public DistributedJobManager(SupervisorProperties conf,
                             SchedJobMapper jobMapper,
                             SchedDependMapper dependMapper,
                             SchedInstanceMapper instanceMapper,
                             SchedTaskMapper taskMapper,
                             SchedWorkflowMapper workflowMapper,
                             cn.ponfee.disjob.common.base.IdGenerator idGenerator,
                             SupervisorRegistry discoveryWorker,
                             TaskDispatcher taskDispatcher,
                             DiscoveryServerRestProxy.GroupedServerInvoker<cn.ponfee.disjob.core.base.WorkerRpcService> groupedWorkerRpcClient,
                             DestinationServerRestProxy.DestinationServerInvoker<cn.ponfee.disjob.core.base.WorkerRpcService,cn.ponfee.disjob.core.base.Worker> destinationWorkerRpcClient,
                             @Qualifier(value="disjobTransactionTemplate")
                             org.springframework.transaction.support.TransactionTemplate transactionTemplate)

方法详细资料

processTaskDispatchFailedEvent

@EventListener
public void processTaskDispatchFailedEvent(TaskDispatchFailedEvent event)

Listen task dispatch failed event

`@Async`需要标注`@EnableAsync`来启用，默认使用的是`SimpleAsyncTaskExecutor`线程池，会为每个任务创建一个新线程(慎用)

updateInstanceNextScanTime

public boolean updateInstanceNextScanTime(cn.ponfee.disjob.core.model.SchedInstance instance,
                                          Date nextScanTime)

savepoint

public boolean savepoint(long taskId,
                         String worker,
                         String executeSnapshot)

abortBroadcastWaitingTask

protected void abortBroadcastWaitingTask(long taskId)

从类复制的说明: AbstractJobManager

Broadcast strategy task after assigned worker. if the worker was dead, should cancel the task.

指定者:

abortBroadcastWaitingTask 在类中 AbstractJobManager

参数:

taskId - the task id

listPausableTasks

protected List<cn.ponfee.disjob.core.model.SchedTask> listPausableTasks(long instanceId)

从类复制的说明: AbstractJobManager

Lists the pausable tasks

指定者:

listPausableTasks 在类中 AbstractJobManager

参数:

instanceId - the instance id

返回:

List

triggerJob

@Transactional(transactionManager="disjobTransactionManager",
               rollbackFor=java.lang.Exception.class)
public void triggerJob(long jobId)
                                                                                                                                    throws cn.ponfee.disjob.core.exception.JobException

Manual trigger the sched job

参数:

jobId - the job id

抛出:

cn.ponfee.disjob.core.exception.JobException - if occur error

createInstance

@Transactional(transactionManager="disjobTransactionManager",
               rollbackFor=java.lang.Exception.class)
public boolean createInstance(cn.ponfee.disjob.core.model.SchedJob job,
                                                                                                                                                  TriggerInstance triggerInstance)

Update sched job, save sched instance and tasks.

参数:

job - the job

triggerInstance - the trigger instance

返回:

true if operated success

updateTaskWorker

@Transactional(transactionManager="disjobTransactionManager",
               rollbackFor=java.lang.Exception.class)
public void updateTaskWorker(String worker,
                                                                                                                                                 List<Long> taskIds)

Set or clear task worker

 当worker不相同时，可使用`CASE WHEN`语法：
 UPDATE sched_task SET worker =
   CASE
     WHEN task_id = 1 THEN 'a'
     WHEN task_id = 2 THEN 'b'
   END
 WHERE id IN (1, 2)
 

参数:

worker - the worker

taskIds - the task id list

startTask

public cn.ponfee.disjob.core.dto.supervisor.StartTaskResult startTask(cn.ponfee.disjob.core.dto.supervisor.StartTaskParam param)

Starts the task

参数:

param - the start task param

返回:

StartTaskResult, if not null start successfully

changeInstanceState

public void changeInstanceState(long instanceId,
                                cn.ponfee.disjob.core.enums.ExecuteState toExecuteState)

Force change instance state

参数:

instanceId - the instance id, unsupported workflow

toExecuteState - the target execute state

deleteInstance

public void deleteInstance(long instanceId)

stopTask

public boolean stopTask(cn.ponfee.disjob.core.dto.supervisor.StopTaskParam param)

Stop task

参数:

param - the stop task param

返回:

true if stopped task successful

purgeInstance

public boolean purgeInstance(cn.ponfee.disjob.core.model.SchedInstance inst)

Purge the zombie instance which maybe dead

参数:

inst - the sched instance

返回:

true if purged successfully

pauseInstance

public boolean pauseInstance(long instanceId)

Pause instance

参数:

instanceId - the instance id, if workflow must be lead instance id

返回:

true if paused successfully

cancelInstance

public boolean cancelInstance(long instanceId,
                              cn.ponfee.disjob.core.enums.Operation ops)

Cancel instance

参数:

instanceId - the instance id, if workflow must be lead instance id

ops - the operation

返回:

true if canceled successfully

resumeInstance

public boolean resumeInstance(long instanceId)

Resume the instance from paused to waiting state

参数:

instanceId - the instance id, if workflow must be lead instance id

返回:

true if resumed successfully