com.netflix.conductor.client.task

Class WorkflowTaskCoordinator

java.lang.Object

com.netflix.conductor.client.task.WorkflowTaskCoordinator

public class WorkflowTaskCoordinator
extends java.lang.Object

Manages the Task workers thread pool and server communication (poll, task update and acknowledgement).

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	WorkflowTaskCoordinator.Builder Builder used to create the instances of WorkflowTaskCoordinator

Constructor Summary

Constructors

Constructor and Description
WorkflowTaskCoordinator(com.netflix.discovery.EurekaClient eurekaClient, TaskClient taskClient, int threadCount, int sleepWhenRetry, int updateRetryCount, int workerQueueSize, java.lang.Iterable<Worker> taskWorkers, java.lang.String workerNamePrefix)

Method Summary

Modifier and Type	Method and Description
int	getSleepWhenRetry()
int	getThreadCount()
int	getUpdateRetryCount()
java.lang.String	getWorkerNamePrefix()
int	getWorkerQueueSize()
void	init() Starts the polling.
void	shutdown()

Methods inherited from class java.lang.Object

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

Constructor Detail

WorkflowTaskCoordinator

public WorkflowTaskCoordinator(com.netflix.discovery.EurekaClient eurekaClient,
                               TaskClient taskClient,
                               int threadCount,
                               int sleepWhenRetry,
                               int updateRetryCount,
                               int workerQueueSize,
                               java.lang.Iterable<Worker> taskWorkers,
                               java.lang.String workerNamePrefix)

Parameters:

eurekaClient - Eureka client - used to identify if the server is in discovery or not. When the server goes out of discovery, the polling is terminated. If passed null, discovery check is not done.

taskClient - TaskClient used to communicate to the Conductor server

threadCount - # of threads assigned to the workers. Should be at-least the size of taskWorkers to avoid starvation in a busy system.

sleepWhenRetry - sleep time in millisecond for Conductor server retries (poll, ack, update task)

updateRetryCount - number of times to retry the failed updateTask operation

workerQueueSize - queue size for the polled task.

taskWorkers - workers that will be used for polling work and task execution.

workerNamePrefix - String prefix that will be used for all the workers.

Please see init() method. The method must be called after this constructor for the polling to start.

See Also:

WorkflowTaskCoordinator.Builder

Method Detail

init

public void init()

Starts the polling. Must be called after the constructor WorkflowTaskCoordinator(EurekaClient, TaskClient, int, int, int, int, Iterable, String) or the builder WorkflowTaskCoordinator.Builder.build() method

shutdown

public void shutdown()

getThreadCount

public int getThreadCount()

Returns:

Thread Count for the executor pool

getWorkerQueueSize

public int getWorkerQueueSize()

Returns:

Size of the queue used by the executor pool

getSleepWhenRetry

public int getSleepWhenRetry()

Returns:

sleep time in millisecond before task update retry is done when receiving error from the Conductor server

getUpdateRetryCount

public int getUpdateRetryCount()

Returns:

Number of times updateTask should be retried when receiving error from Conductor server

getWorkerNamePrefix

public java.lang.String getWorkerNamePrefix()

Returns:

prefix used for worker names