| Package | Description |
|---|---|
| org.apache.drill.yarn.appMaster |
Implements the Drill Application Master for YARN.
|
| org.apache.drill.yarn.zk |
Interface between the Application Master and ZooKeeper.
|
| Modifier and Type | Method and Description |
|---|---|
void |
Scheduler.TaskManager.allocated(EventContext context) |
void |
AbstractDrillbitScheduler.DrillbitManager.allocated(EventContext context) |
void |
AbstractTaskManager.allocated(EventContext context) |
void |
TaskState.cancel(EventContext context)
Cluster manager wishes to cancel this task.
|
void |
TaskState.EndingState.cancel(EventContext context) |
void |
TaskState.KillingState.cancel(EventContext context) |
void |
SchedulerStateActions.checkTasks(EventContext context,
long curTime) |
void |
SchedulerStateImpl.checkTasks(EventContext context,
long curTime) |
void |
Scheduler.TaskManager.completed(EventContext context) |
void |
AbstractDrillbitScheduler.DrillbitManager.completed(EventContext context) |
void |
AbstractTaskManager.completed(EventContext context) |
protected void |
TaskState.completed(EventContext context,
org.apache.hadoop.yarn.api.records.ContainerStatus status) |
void |
TaskState.completionAck(EventContext context)
The monitoring plugin has detected that the task has confirmed that it has
started shutdown.
|
void |
TaskState.EndingState.completionAck(EventContext context) |
void |
TaskState.KillingState.completionAck(EventContext context) |
void |
SchedulerStateActions.containerAllocated(EventContext context,
org.apache.hadoop.yarn.api.records.Container container)
A container request has been granted.
|
void |
SchedulerStateImpl.containerAllocated(EventContext context,
org.apache.hadoop.yarn.api.records.Container container) |
void |
TaskState.containerAllocated(EventContext context,
org.apache.hadoop.yarn.api.records.Container container)
Resource Manager reports that the task has been allocated a container.
|
void |
TaskState.containerCompleted(EventContext context,
org.apache.hadoop.yarn.api.records.ContainerStatus status)
Resource Manager reports that the task has completed execution and provided
the completion status.
|
void |
TaskState.EndingState.containerCompleted(EventContext context,
org.apache.hadoop.yarn.api.records.ContainerStatus status)
Normal ENDING --> WAIT_END_ACK | END transition.
|
void |
TaskState.KillingState.containerCompleted(EventContext context,
org.apache.hadoop.yarn.api.records.ContainerStatus status)
Normal KILLING --> WAIT_END_ACK | END transition.
|
void |
TaskState.containerStarted(EventContext context)
Node Manager reports that the task has started execution.
|
void |
TaskState.containerStopped(EventContext context)
Node Manager reports that the task has stopped execution.
|
protected void |
TaskState.endOrAck(EventContext context) |
void |
ClusterControllerImpl.fireLifecycleChange(TaskLifecycleListener.Event event,
EventContext context) |
boolean |
Scheduler.TaskManager.isLive(EventContext context) |
boolean |
AbstractDrillbitScheduler.DrillbitManager.isLive(EventContext context) |
boolean |
AbstractTaskManager.isLive(EventContext context) |
void |
TaskState.launchFailed(EventContext context,
Throwable t)
The launch of the container failed.
|
void |
TaskState.requestContainer(EventContext context) |
boolean |
SchedulerStateActions.requestContainers(EventContext context,
int maxRequests)
Request a container the first task that we wish to start.
|
boolean |
SchedulerStateImpl.requestContainers(EventContext context,
int maxRequests) |
void |
SchedulerStateActions.shutDown(EventContext context)
Shut down this task group by canceling all tasks not already cancelled.
|
void |
SchedulerStateImpl.shutDown(EventContext context) |
void |
TaskState.startAck(EventContext context)
The monitoring plugin has detected that the task has confirmed that it is
fully started.
|
void |
TaskState.KillingState.startAck(EventContext context) |
void |
TaskLifecycleListener.stateChange(TaskLifecycleListener.Event event,
EventContext context) |
void |
TaskState.stopTaskFailed(EventContext context,
Throwable t)
The node manager request to stop a task failed.
|
void |
TaskState.KillingState.stopTaskFailed(EventContext context,
Throwable t) |
protected void |
TaskState.taskStartFailed(EventContext context,
Task.Disposition disposition)
Task failed when starting.
|
protected void |
TaskState.taskTerminated(EventContext context)
A running task terminated.
|
void |
TaskState.tick(EventContext context,
long curTime) |
void |
TaskState.EndingState.tick(EventContext context,
long curTime)
If the graceful stop process exceeds the maximum timeout, go ahead and
forcibly kill the process.
|
protected void |
TaskState.transition(EventContext context,
TaskState newState)
Implement a state transition, alerting any life cycle listeners and
updating the log file.
|
| Modifier and Type | Method and Description |
|---|---|
void |
ZKRegistry.stateChange(TaskLifecycleListener.Event event,
EventContext context)
Listen for selected YARN task state changes.
|
Copyright © 2021 The Apache Software Foundation. All rights reserved.