hydra

Class ClientNameTaskScheduler

java.lang.Object

hydra.BaseTaskScheduler

hydra.ClientNameTaskScheduler

All Implemented Interfaces:

TaskScheduler

Direct Known Subclasses:

ClientNameLockStepTaskScheduler

public class ClientNameTaskScheduler
extends BaseTaskScheduler

This class acts as the task scheduler for tasks that run at the start and end of a test, in single-threaded VMs started just to run the tasks, that is, tasks of type STARTTASK and ENDTASK.

The set of client threads runs tasks concurrently, but each client thread executes each task in sequence. One client thread can get ahead of the other in its sequence.

Sample timeline with 3 clients running 3 tasks...

  c1:  t1---------------t2------t3------|
  c2:  t1----t2------t3---|
  c3:  t1---------t2---------t3----------|
  

Each task is executed only in VMs with the logical client names specified for that task in the test configuration file, or, if none are specified, in the VM for each and every logical client name defined in the file. See ClientPrms.names for how to specify clients.

Field Summary

Fields inherited from class hydra.BaseTaskScheduler

disconnect, haltIfBadResult, log, name, purpose, tasks, termination, vms

Constructor Summary

Constructors

Constructor and Description
ClientNameTaskScheduler(String name, Vector tasks, String purpose) Creates a task scheduler by the given name for the purpose of executing the given set of start or end tasks.

Method Summary

Modifier and Type	Method and Description
void	createClients(Map cds, Map tgs) Asks this scheduler to create one single-threaded client vm for each client description required by the tasks, rather than using existing clients.
protected boolean	executeTasks() Executes the tasks using the existing clients.

Methods inherited from class hydra.BaseTaskScheduler

assignTask, exceededMaxResultWaitSec, exceededMaxResultWaitSec, exceededMaxResultWaitSec, executedNormally, executedNormallyOrNoHaltOrder, executeTasks, fakeKillResults, getTsid, getVm, getVms, hadDynamicActionError, hadDynamicActionHang, hadFatalTaskErrorAndHaltOrder, hadStatMonitorError, initClientRuntime, initExecutionParameters, isFinished, pauseBlocking, pauseScheduling, pauseWaiting, printBlockingReport, printReport, printSchedulingReport, printWaitingReport, processDynamicActionError, reportResult, schedulerResult, setDynamicActionError, setDynamicActionHang, setStatMonitorError, setTermination, startClients, startClientsForced, stopClients, waitForClientsToFinish, waitForClientsToFinish, waitForClientToFinish

Methods inherited from class java.lang.Object

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

Constructor Detail

ClientNameTaskScheduler

public ClientNameTaskScheduler(String name,
                               Vector tasks,
                               String purpose)

Creates a task scheduler by the given name for the purpose of executing the given set of start or end tasks.

Parameters:

name - a descriptive name of the tasks to use for logging

tasks - the tasks to be scheduled

Method Detail

createClients

public void createClients(Map cds,
                          Map tgs)

Asks this scheduler to create one single-threaded client vm for each client description required by the tasks, rather than using existing clients.

Specified by:

createClients in interface TaskScheduler

Overrides:

createClients in class BaseTaskScheduler

Parameters:

cds - the ClientDescription describes the client vms and threads

tgs - the thread groups to map clients to.

Throws:

HydraTimeoutException - if the clients do not start within Prms.maxClientStartupWaitSec

executeTasks

protected boolean executeTasks()

Executes the tasks using the existing clients.

Specified by:

executeTasks in class BaseTaskScheduler

Returns:

true if there were no fatal task errors or hangs