Package org.apache.ignite.compute

Interface ComputeTaskContinuousMapper

public interface ComputeTaskContinuousMapper

Defines a mapper that can be used for asynchronous job sending. Useful for streaming jobs within the same task. Note that if job number within a task grows too large, it is best to attach ComputeTaskNoResultCache annotation to task to make sure that collection of job results and job siblings does not grow indefinitely.

Continuous mapper methods can be used right after it injected into a task. Mapper can not be used after ComputeTask.result(ComputeJobResult, List) method returned the ComputeJobResultPolicy.REDUCE policy. Also if ComputeTask.result(ComputeJobResult, List) method returned the ComputeJobResultPolicy.WAIT policy and all jobs are finished then task will go to reducing results and continuous mapper can not be used.

Note that whenever continuous mapper is used, ComputeTask.map(List, Object) method is allowed to return null in case when at least one job has been sent prior to completing the ComputeTask.map(List, Object) method.

Task continuous mapper can be injected into a task using IoC (dependency injection) by attaching TaskContinuousMapperResource annotation to a field or a setter method inside of ComputeTask implementations as follows:

 ...
 // This field will be injected with task continuous mapper.
 @TaskContinuousMapperResource
 private ComputeTaskContinuousMapper mapper;
 ...
 

or from a setter method:

 // This setter method will be automatically called by the system
 // to set grid task continuous mapper.
 @TaskContinuousMapperResource
 void setSession(ComputeTaskContinuousMapper mapper) {
     this.mapper = mapper;
 }
 

Method Summary

Modifier and Type	Method	Description
void	send(Collection<? extends ComputeJob> jobs)	Sends collection of jobs to nodes automatically picked by the underlying load balancer.
void	send(Map<? extends ComputeJob,ClusterNode> mappedJobs)	Sends collection of grid jobs to assigned nodes.
void	send(ComputeJob job)	Sends job to a node automatically picked by the underlying load balancer.
void	send(ComputeJob job, ClusterNode node)	Sends given job to a specific grid node.

Method Detail

send

void send(ComputeJob job,
          ClusterNode node)
   throws IgniteException

Sends given job to a specific grid node.

Parameters:

job - Job instance to send. If null is passed, exception will be thrown.

node - Grid node. If null is passed, exception will be thrown.

Throws:

IgniteException - If job can not be processed.

send

void send(Map<? extends ComputeJob,ClusterNode> mappedJobs)
   throws IgniteException

Sends collection of grid jobs to assigned nodes.

Parameters:

mappedJobs - Map of grid jobs assigned to grid node. If null or empty list is passed, exception will be thrown.

Throws:

IgniteException - If job can not be processed.

send

void send(ComputeJob job)
   throws IgniteException

Sends job to a node automatically picked by the underlying load balancer.

Parameters:

job - Job instance to send. If null is passed, exception will be thrown.

Throws:

IgniteException - If job can not be processed.

send

void send(Collection<? extends ComputeJob> jobs)
   throws IgniteException

Sends collection of jobs to nodes automatically picked by the underlying load balancer.

Parameters:

jobs - Collection of grid job instances. If null or empty list is passed, exception will be thrown.

Throws:

IgniteException - If job can not be processed.