def status(simfolder):
"""
Check status of simulation. If on PBS scheduling, show status of nodes.
Then, ask the fjd-dispatcher about status of jobs.
:param string simfolder: relative path to simfolder
:returns: True if successful, False otherwise
"""
scheduler = utils.get_scheduler(simfolder)
sim_name = utils.get_simulation_name(simfolder, "{}/stosim.conf".format(simfolder))
if scheduler == 'pbs':
num_nodes = len([n for n in os.listdir('{}/jobs'.format(simfolder))\
if n.endswith('.pbs')])
if num_nodes > 0:
print("[StoSim] State of our {} PBS computing nodes:".format(num_nodes))
subprocess.call('echo "Waiting: $(qselect -u $USER -s W | wc -l)"', shell=True)
subprocess.call('echo "Queued: $(qselect -u $USER -s Q | wc -l)"', shell=True)
subprocess.call('echo "Running: $(qselect -u $USER -s R | wc -l)"', shell=True)
else:
print("[StoSim] No PBS computing nodes seem to be configured...")
print("[StoSim] State of workers and jobs:")
subprocess.call('fjd-dispatcher --project {} --status_only --interval {}'\
.format(sim_name, utils.get_interval(simfolder)), shell=True)
return True
def resume(simfolder):
"""
(Re)start dispatching jobs
:param string simfolder: relative path to simfolder
:returns: True if successful, False otherwise
"""
sim_name = utils.get_simulation_name(simfolder, "{}/stosim.conf".format(simfolder))
subprocess.call('fjd-dispatcher --project {} --interval {}'\
.format(sim_name, utils.get_interval(simfolder)), shell=True)
return True
def run(simfolder):
''' The main function to start running simulations
:param string simfolder: relative path to simfolder
:returns: True if successful, False otherwise
'''
print('*' * 80)
sim_name = utils.get_simulation_name(simfolder, "{}/stosim.conf".format(simfolder))
print("Running simulation {}".format(sim_name))
print('*' * 80)
print('')
if not osp.exists("%s/stosim.conf" % simfolder):
print("[StoSim] %s/stosim.conf does not exist!" % simfolder)
utils.usage()
return False
# prepare all jobs to be run by FJD
fjd_dir = fjd.utils.ensure_wdir(sim_name)
fjd.utils.empty_queues(sim_name)
for job in [j for j in os.listdir("{}/jobs".format(simfolder))\
if j.endswith('.conf')]:
copy("{}/jobs/{}".format(simfolder, job),
"{}/jobqueue".format(fjd_dir))
dispatch_cmd = 'fjd-dispatcher --project {} --end_when_jobs_are_done '\
' --callback "stosim --kill" --interval {}'\
.format(sim_name, utils.get_interval(simfolder))
# now decide if recruiting is done in a local network or on a PBS cluster
scheduler = utils.get_scheduler(simfolder)
if scheduler == 'fjd':
# let FJD handle it in local network (default: only local PC)
if os.path.exists('{}/remote.conf'.format(simfolder)):
copy('{}/remote.conf'.format(simfolder), fjd_dir)
subprocess.call('fjd-recruiter --project {} hire'.format(sim_name), shell=True)
if not we_exited[0]:
subprocess.call(dispatch_cmd, shell=True)
# when recruiter got remote.conf, clean up in fjd dir
if os.path.exists('{}/remote.conf'.format(simfolder)):
os.remove('{}/remote.conf'.format(fjd_dir))
elif scheduler == 'pbs':
# queue the PBS jobs we created on a PBS job scheduler (e.g. clusters
# running Torque or PBS Pro). These simply start FJD workers.
for job in [j for j in os.listdir('{}/jobs'.format(simfolder)) if j.endswith('.pbs')]:
subprocess.call('qsub {}'.format('{}/jobs/{}'.format(simfolder, job)), shell=True)
# Now we start dispatching
subprocess.call(dispatch_cmd, shell=True)
return True