# Create a new resource handle with one resource and a fixed # number of cores and runtime. cluster = ResourceHandle( resource=resource, cores=config[resource]["cores"], walltime=15, #username=None, project=config[resource]['project'], access_schema=config[resource]['schema'], queue=config[resource]['queue'], #database_url='mongodb://138.201.86.166:27017/ee_exp_4c', ) # Allocate the resources. cluster.allocate() # Set the 'instances' of the pipeline to 16. This means that 16 instances # of each pipeline stage are executed. # # Execution of the 16 pipeline instances can happen concurrently or # sequentially, depending on the resources (cores) available in the # SingleClusterEnvironment. ccount = RunExchange(stages=3, instances=2) cluster.run(ccount) except EnsemblemdError, er: print "Ensemble MD Toolkit Error: {0}".format(str(er)) raise # Just raise the execption again to get the backtrace
# Create a new resource hande with one resource and a fixed # number of cores and runtime. cluster = ResourceHandle( resource=resource, cores=config[resource]["cores"], walltime=15, #username=None, project=config[resource]['project'], access_schema = config[resource]['schema'], queue = config[resource]['queue'], database_url='mongodb://*****:*****@ds015335.mlab.com:15335/rp', ) # Allocate the resources. cluster.allocate() # We set the simulation 'instances' to 16 and analysis 'instances' to 1. We set the adaptive # simulation to True and specify the simulation extraction script to be used. cur_path = os.path.dirname(os.path.abspath(__file__)) mssa = MSSA(iterations=2, simulation_instances=16, analysis_instances=1, adaptive_simulation=True, sim_extraction_script='{0}/extract.py'.format(cur_path)) cluster.run(mssa) except EnsemblemdError, er: print "Ensemble MD Toolkit Error: {0}".format(str(er)) raise # Just raise the execption again to get the backtrace try: cluster.deallocate()
pipe = Test(ensemble_size=ENSEMBLE_SIZE + 1, pipeline_size=1) # Create an application manager app = AppManager(name='Adap_sampling') # Register kernels to be used app.register_kernels(rand_kernel) app.register_kernels(sleep_kernel) # Add workload to the application manager app.add_workload(pipe) # Create a resource handle for target machine res = ResourceHandle( resource="local.localhost", cores=4, # username=, # project =, # queue=, walltime=10, database_url='mongodb://ensembletk.imp.fu-berlin.de:27017/rp') # Submit request for resources + wait till job becomes Active res.allocate(wait=True) # Run the given workload res.run(app) # Deallocate the resource res.deallocate()