org.openspaces.grid.gsm.machines.plugins
Interface NonBlockingElasticMachineProvisioning

All Superinterfaces:

ElasticGridServiceAgentProvisioningProgressEventListenerAware, ElasticMachineProvisioningProgressChangedEventListenerAware, ElasticMachineProvisioningSpaceAware, ElasticProcessingUnitMachineIsolationAware

All Known Implementing Classes:

DiscoveredMachineProvisioningBean, NonBlockingElasticMachineProvisioningAdapter

public interface NonBlockingElasticMachineProvisioning
extends ElasticProcessingUnitMachineIsolationAware, ElasticMachineProvisioningSpaceAware, ElasticMachineProvisioningProgressChangedEventListenerAware, ElasticGridServiceAgentProvisioningProgressEventListenerAware

An Elastic Service Manager plug-in that enables starting, stopping and discovering of virtual machines. The plugin must be stateless, and is expected to delegate calls to a remote service (such as a cloud). Calls to this class are guaranteed to be called from the same thread.

Since:

8.0

Author:

itaif

See Also:

ElasticMachineProvisioning, NonBlockingElasticMachineProvisioningAdapter, ElasticMachineProvisioningConfig

Method Summary

FutureCleanupCloudResources	cleanupCloudResources(long duration, TimeUnit unit) Cleanup all cloud resources.
ElasticMachineProvisioningConfig	getConfig()
FutureGridServiceAgents	getDiscoveredMachinesAsync(long duration, TimeUnit unit)
boolean	isStartMachineSupported()
FutureMachineFailure[]	onMachinesFailureAsync(FailedGridServiceAgent[] failedAgents, long duration, TimeUnit unit) Perform the required actions following an unexpected machine failure.
FutureGridServiceAgent[]	startMachinesAsync(CapacityRequirements capacityRequirements, ExactZonesConfig zones, FailedGridServiceAgent[] failedAgents, long duration, TimeUnit unit) Starts a new machine with a new grid service agent with the specified grid service agent zones.
FutureStoppedMachine	stopMachineAsync(GridServiceAgent agent, long duration, TimeUnit unit) Shuts down the grid service agent and the machine.

Methods inherited from interface org.openspaces.grid.gsm.machines.isolation.ElasticProcessingUnitMachineIsolationAware

setElasticProcessingUnitMachineIsolation

Methods inherited from interface org.openspaces.grid.gsm.machines.plugins.space.ElasticMachineProvisioningSpaceAware

setElasticMachineProvisioningSpace

Methods inherited from interface org.openspaces.grid.gsm.machines.plugins.events.ElasticMachineProvisioningProgressChangedEventListenerAware

setElasticMachineProvisioningProgressChangedEventListener

Methods inherited from interface org.openspaces.grid.gsm.machines.plugins.events.ElasticGridServiceAgentProvisioningProgressEventListenerAware

setElasticGridServiceAgentProvisioningProgressEventListener

Method Detail

isStartMachineSupported

boolean isStartMachineSupported()

Returns:

true if this object supports starting and stopping of machines. False indicates that #startMachinesAsync(CapacityRequirements, long, TimeUnit) and stopMachineAsync(GridServiceAgent, long, TimeUnit) raise UnsupportedOperationException

Since:

8.0.1

getDiscoveredMachinesAsync

FutureGridServiceAgents getDiscoveredMachinesAsync(long duration,
                                                   TimeUnit unit)

Parameters:

duration - - the maximum duration after which a TimeoutException is raised.

unit - - the time unit for the duration

Returns:

a future array of grid service agents that reside in the data center (or cloud region) defined in ElasticMachineProvisioningConfig The caller then filters the result using the criteria defined in ElasticMachineProvisioningConfig.getGridServiceAgentZones() , ElasticMachineProvisioningConfig.isGridServiceAgentZoneMandatory() and ElasticMachineProvisioningConfig.isDedicatedManagementMachines() so it is not mandatory for the implementation to do so.

Since:

8.0.1

getConfig

ElasticMachineProvisioningConfig getConfig()

Returns:

the configuration used by this object

Since:

8.0.1

startMachinesAsync

FutureGridServiceAgent[] startMachinesAsync(CapacityRequirements capacityRequirements,
                                            ExactZonesConfig zones,
                                            FailedGridServiceAgent[] failedAgents,
                                            long duration,
                                            TimeUnit unit)

Starts a new machine with a new grid service agent with the specified grid service agent zones. This method is non blocking and returns a future object with the new grid service agent.

Parameters:

capacityRequirements - - total capacity of new machines

zones - - the agent zones.

duration - - the maximum duration after which a TimeoutException is raised.

unit - - the time unit for the duration.

failedAgents - - list of agents that require recovery

Returns:

the grid service agent futures

Throws:

ElasticMachineProvisioningException

InterruptedException

TimeoutException

Since:

9.1.0

stopMachineAsync

FutureStoppedMachine stopMachineAsync(GridServiceAgent agent,
                                      long duration,
                                      TimeUnit unit)

Shuts down the grid service agent and the machine. The implementation should be able to close machines that it has not started, but rather an older instance of this object started perhaps with different configuration. This method is non blocking and is idempotent.

Parameters:

agent -

duration -

unit -

Returns:

Throws:

TimeoutException

InterruptedException

ElasticMachineProvisioningException

Since:

8.0

cleanupCloudResources

FutureCleanupCloudResources cleanupCloudResources(long duration,
                                                  TimeUnit unit)

Cleanup all cloud resources. This method is called once after the processing unit has undeployed and all machines have been stopped.

Since:

9.7.0

onMachinesFailureAsync

FutureMachineFailure[] onMachinesFailureAsync(FailedGridServiceAgent[] failedAgents,
                                              long duration,
                                              TimeUnit unit)

Perform the required actions following an unexpected machine failure. E.g. handle machine related resources (e.g. floating IPs, attached storage). This method is called once during failover, just before starting a new machine. The method is non blocking and returns an array of future objects, one for each failed machine that was handled.

Parameters:

failedAgents - - list of agents that require recovery

Returns:

an array of future objects, one for each failed machine that was handled

Throws:

ElasticMachineProvisioningException

Since:

9.7.1