def _setup_flowsheet_turbine_node(dat, nkey, user_name):
""" From s3 download all simulation files into AspenSinterComsumer cache directory '{working_directory\test\{simulation_guid}'. If
Simulation does not exist create one. If Simulation does exist just s3 download all simulation files into the above cache directory.
The new simulation_guid is created for all file updates to TurbineWS, so this is sidestepping that process.
TODO: Provide a simulation_id via S3 ( eg. {simulation_name}/Id )
"""
assert len(dat.flowsheet.nodes[nkey].turbApp) == 2, \
'DAT Flowsheet nodes turbApp is %s' %dat.flowsheet.nodes[nkey].turbApp
node = dat.flowsheet.nodes[nkey]
turb_app = node.turbApp[0]
model_name = node.modelName
assert turb_app is not None
turb_app = turb_app.lower()
assert turb_app in ['acm', 'aspenplus'], 'unknown turbine application "%s"' %turb_app
""" Search S3 Bucket for node simulation
"""
s3 = boto3.client('s3', region_name=AWS_REGION)
bucket_name = FOQUSAWSConfig.get_instance().get_simulation_bucket_name()
prefix = '%s/%s/' %(user_name,model_name)
l = s3.list_objects(Bucket=bucket_name, Prefix=prefix)
assert 'Contents' in l, 'Node %s failure: S3 Bucket %s is missing simulation files for "%s"' %(nkey, bucket_name, prefix)
key_sinter_filename = None
key_model_filename = None
s3_key_list = [i['Key'] for i in l['Contents']]
_log.debug('Node model %s staged-input files %s' %(model_name, s3_key_list))
for k in s3_key_list:
if k.endswith('/%s_sinter.json' %turb_app):
key_sinter_filename = k
elif turb_app == 'acm' and k.endswith('.acmf'):
assert key_model_filename is None, 'detected multiple model files'
key_model_filename = k
elif turb_app == 'aspenplus' and k.endswith('.bkp'):
assert key_model_filename is None, 'detected multiple model files'
key_model_filename = k
assert key_sinter_filename is not None, 'Flowsheet node=%s simulation=%s sinter configuration not in %s' %(nkey, model_name, str(s3_key_list))
assert key_model_filename is not None, 'Flowsheet node=%s simulation=%s model file not in %s' %(nkey, model_name, str(s3_key_list))
""" search TurbineLite WS for node simulation
"""
print(turbine_simulation_script.__file__)
turbine_cfg = node.gr.turbConfig.getFile()
_log.debug('CWD: %s', os.path.abspath(os.path.curdir))
turbine_cfg = os.path.abspath(turbine_cfg)
_log.debug('Turbine Configuration File: %s', turbine_cfg)
sim_list = turbine_simulation_script.main_list([turbine_cfg], func=None)
print('Simulation List %s' %sim_list)
sim_d = [i for i in sim_list if i['Name'] == model_name]
cache_sim_guid = None
assert len(sim_d) < 2, 'Expecting 0 or 1 entries for simulation %s' %model_name
if len(sim_d) == 0:
_log.debug('No simulation="%s" in TurbineLite' %model_name)
sim_d = None
cache_sim_guid = str(uuid.uuid4())
else:
_log.debug('Found simulation="%s" in TurbineLite' %model_name)
sim_d = sim_d[0]
assert 'Id' in sim_d, 'Missing keys in Simulation %s' %sim_d
cache_sim_guid = sim_d['Id']
""" upload all staged-inputs to TurbineLite if new or updated in
s3://{bucketname}/{username}/{simulation}
"""
entry_list = [i for i in l['Contents'] if i['Key'] != prefix and i['Key'].startswith(prefix)]
update_required = False
#target_dir = os.path.join(CURRENT_JOB_DIR, model_name)
target_dir = os.path.join(WORKING_DIRECTORY, 'test', cache_sim_guid)
os.makedirs(target_dir, exist_ok=True)
sinter_local_filename = None
for entry in entry_list:
_log.debug("s3 staged input: %s", entry)
key = entry['Key']
etag = entry.get('ETag', "").strip('"')
# Upload to TurbineLite
# if ends with json or acmf
si_metadata = []
target_file_path = None
assert key.startswith(prefix)
if key == key_sinter_filename:
#assert key_sinter_filename == '/'.join(prefix, key_sinter_filename.split('/')[-1]), \
# 'sinter configuration "%s" must be in model base directory: "%s"' %(key_model_filename,prefix)
target_file_path = os.path.join(target_dir, "sinter_configuration.txt")
sinter_local_filename = target_file_path
if sim_d: si_metadata = [i for i in sim_d["StagedInputs"] if i['Name'] == 'configuration']
s3.download_file(bucket_name, key, target_file_path)
elif key == key_model_filename:
#assert key_model_filename == '/'.join(prefix, key_model_filename.split('/')[-1]), \
# 'sinter configuration "%s" must be in model base directory: "%s"' %(key_model_filename,prefix)
target_file_path = os.path.join(target_dir, key.split('/')[-1])
if sim_d: si_metadata = [i for i in sim_d["StagedInputs"] if i['Name'] == 'aspenfile']
s3.download_file(bucket_name, key, target_file_path)
else:
args = [ i for i in key[len(prefix):].split('/') if i ]
args.insert(0, target_dir)
target_file_path = os.path.join(*args)
if sim_d: si_metadata = [i for i in sim_d["StagedInputs"] if i['Name'] == key.split('/')[-1]]
s3.download_file(bucket_name, key, target_file_path)
_log.debug('model="%s" key="%s" staged-in file="%s"' %(model_name, key, target_file_path))
assert len(si_metadata) < 2, 'Turbine Error: Duplicate entries for "%s", "%s"' %(model_name, key)
"""NOTE: Multipart uploads have different ETags ( end with -2 or something )
Thus the has comparison will fail. For now ignore it, but fixing this check is performance optimization.
if len(si_metadata) == 1:
file_hash = si_metadata[0]['MD5Sum']
if file_hash.lower() != etag.lower():
_log.debug('Updated detected(hash "%s" != "%s"): s3.getObject "%s"' %(etag,file_hash,key))
s3.download_file(bucket_name, key, target_file_path)
update_required = True
else:
_log.debug('md5 matches for staged-in file "%s"' %key)
else:
_log.debug('Add to Turbine Simulation(%s) s3.getObject: "%s"' %(model_name, key))
s3.download_file(bucket_name, key, target_file_path)
update_required = True
"""
assert sinter_local_filename is not None, 'missing sinter configuration file'
if sim_d is None:
_log.debug('Adding Simulation "%s" "%s"' %(model_name,cache_sim_guid))
node.gr.turbConfig.uploadSimulation(model_name, sinter_local_filename, guid=cache_sim_guid, update=False)
"""
def test_GET_List(self):
l = tss.main_list([self.config_name], func=None)
self.log.debug('simulation names %s' %l)
self.failUnless(type(l) is list,'return type should be list')
def test_GET_List(self):
simulation_list = tss.main_list([self.config_name], func=None)
simulation_names = map(lambda i: i['Name'], simulation_list)
self.log.debug('simulation names %s' %simulation_names)
self.failUnless(set([]).issubset(simulation_names),
'%s not superset' %simulation_names)
def setup_foqus(self, db, job_desc):
"""
Move job to state setup
Pull FOQUS nodes' simulation files from AWS S3
ACM simulations store in TurbineLite
"""
sfile,rfile,vfile,ofile = getfilenames(job_desc['Id'])
guid = job_desc['Id']
jid = None
simId = job_desc['Simulation']
# Run the job
db.add_message("consumer={0}, starting job {1}"\
.format(db.consumer_id, jid), guid)
_log.debug("setup foqus")
db.job_change_status(guid, "setup")
configContent = db.get_configuration_file(simId)
logging.getLogger("foqus." + __name__)\
.info("Job {0} is submitted".format(jid))
#db.jobConsumerID(guid, consumer_uuid)
#db.job_prepare(guid, jid, configContent)
# Load the session file
dat = Session(useCurrentWorkingDir=True)
dat.load(sfile, stopConsumers=True)
dat.loadFlowsheetValues(vfile)
# dat.flowsheet.nodes.
s3 = boto3.client('s3', region_name='us-east-1')
bucket_name = 'foqus-simulations'
flowsheet_name = job_desc['Simulation']
username = '******'
prefix = '%s/%s' %(username,flowsheet_name)
l = s3.list_objects(Bucket=bucket_name, Prefix=prefix)
assert l.has_key('Contents'), "S3 Simulation: No keys match %s" %prefix
_log.debug("Process Flowsheet nodes")
for nkey in dat.flowsheet.nodes:
if dat.flowsheet.nodes[nkey].turbApp is None:
continue
assert len(dat.flowsheet.nodes[nkey].turbApp) == 2, \
'DAT Flowsheet nodes turbApp is %s' %dat.flowsheet.nodes[nkey].turbApp
node = dat.flowsheet.nodes[nkey]
turb_app = node.turbApp[0]
model_name = node.modelName
#sinter_filename = 'anonymous/%s/%s/%s.json' %(job_desc['Simulation'],nkey, model_name)
sinter_filename = '/'.join((username, flowsheet_name, nkey, '%s.json' %model_name))
s3_key_list = map(lambda i: i['Key'] , l['Contents'])
assert sinter_filename in s3_key_list, 'missing sinter configuration "%s" not in %s' %(sinter_filename, str(s3_key_list))
simulation_name = job_desc.get('Simulation')
#sim_list = node.gr.turbConfig.getSimulationList()
sim_list = turbine_simulation_script.main_list([node.gr.turbConfig.getFile()])
_log.info("Node Turbine Simulation Requested: (%s, %s)", turb_app, simulation_name)
if turb_app == 'ACM':
model_filename = 'anonymous/%s/%s/%s.acmf' %(simulation_name,nkey, model_name)
assert model_filename in s3_key_list, 'missing sinter configuration "%s"' %sinter_filename
else:
_log.info("Turbine Application Not Implemented: '%s'", turb_app)
raise NotImplementedError, 'Flowsheet Node model type: "%s"' %(str(dat.flowsheet.nodes[nkey].turbApp))
sim_d = filter(lambda i: i['Name'] == model_name, sim_list)
assert len(sim_d) < 2, 'Expecting 0 or 1 entries for simulation %s' %simulation_name
if len(sim_d) == 0:
sim_d = None
else:
sim_d = sim_d[0]
prefix = 'anonymous/%s/%s/' %(job_desc['Simulation'],nkey)
entry_list = filter(lambda i: i['Key'] != prefix and i['Key'].startswith(prefix), l['Contents'])
sinter_local_filename = None
update_required = False
for entry in entry_list:
_log.debug("ENTRY: %s", entry)
key = entry['Key']
etag = entry.get('ETag', "").strip('"')
file_name = key.split('/')[-1]
file_path = os.path.join(CURRENT_JOB_DIR, file_name)
# Upload to TurbineLite
# if ends with json or acmf
si_metadata = []
if key.endswith('.json'):
_log.debug('CONFIGURATION FILE')
sinter_local_filename = file_path
if sim_d:
si_metadata = filter(lambda i: i['Name'] == 'configuration', sim_d["StagedInputs"])
elif key.endswith('.acmf'):
_log.debug('ACMF FILE')
if sim_d:
si_metadata = filter(lambda i: i['Name'] == 'aspenfile', sim_d["StagedInputs"])
else:
raise NotImplementedError, 'Not allowing File "%s" to be staged in' %key
assert len(si_metadata) < 2, 'Turbine Error: Too many entries for "%s", "%s"' %(simulation_name, file_name)
# NOTE: Multipart uploads have different ETags ( end with -2 or something )
# THus the has comparison will fail
# FOr now ignore it, but fixing this check is performance optimization.
#
if len(si_metadata) == 1:
file_hash = si_metadata[0]['MD5Sum']
if file_hash.lower() != etag.lower():
_log.debug("Compare %s: %s != %s" %(file_name, etag, file_hash))
_log.debug('s3 download(%s): %s' %(CURRENT_JOB_DIR, key))
s3.download_file(bucket_name, key, file_path)
update_required = True
else:
_log.debug("MATCH")
s3.download_file(bucket_name, key, file_path)
else:
_log.debug("Add to Turbine Simulation(%s) File: %s" %(simulation_name, file_name))
s3.download_file(bucket_name, key, file_path)
update_required = True
assert sinter_local_filename is not None, 'missing sinter configuration file'
if model_name not in map(lambda i: i['Name'], sim_list):
_log.debug('Adding Simulation "%s"' %model_name)
node.gr.turbConfig.uploadSimulation(model_name, sinter_local_filename, update=False)
elif update_required:
# NOTE: Requires the configuration file on update, so must download_file it above...
_log.debug('Updating Simulation "%s"' %model_name)
node.gr.turbConfig.uploadSimulation(model_name, sinter_local_filename, update=True)
else:
_log.debug('No Update Required for Simulation "%s"' %model_name)
return dat
def run_foqus(job_desc):
"""
job_desc: {"Initialize":false,
"Input":{"BFBRGN.Cr":1,"BFBRGN.Dt":9.041,"BFBRGN.Lb":8.886,
"BFBRGNTop.Cr":1,"BFBRGNTop.Dt":9.195,"BFBRGNTop.Lb":7.1926,
"BFBadsB.Cr":1,"BFBadsB.Dt":11.897,"BFBadsB.Lb":2.085,
"BFBadsB.dx":0.0127,"BFBadsM.Cr":1,"BFBadsM.Dt":15,"BFBadsM.Lb":1.972,
"BFBadsM.dx":0.06695,"BFBadsT.Cr":1,"BFBadsT.Dt":15,"BFBadsT.Lb":2.203,
"BFBadsT.dx":0.062397,"GHXfg.A_exch":16358,"GHXfg.GasIn.P":1.01325,
"GHXfg.GasIn.T":54,"Kd":100,"MinStepSize":0.001,"RunMode":"Steady State",
"Script":"Run_Steady","Snapshot":"","TimeSeries":[0],"TimeUnits":"Hours",
"dp":0.00015,"fg_flow":100377,"homotopy":0,"printlevel":0},
"Reset":false,
"Simulation":"BFB_v11_FBS_01_26_2018",
"Visible":false,
"Id":"8a3033b4-6de2-409c-8552-904889929704"}
"""
exit_code = 0
sfile, rfile, vfile, ofile = getfilenames(job_desc['Id'])
# Session file to run
load_gui = False
dat = Session(useCurrentWorkingDir=True)
#Make ctrl-c do nothing but and SIGINT donothing but interupt
#the loop
#signal.signal(signal.SIGINT, signal_handler)
#Register consumer TurbineLite DB
db = TurbineLiteDB()
#db.dbFile = os.path.join(dat.foqusSettings.turbLiteHome,
# "Data/TurbineCompactDatabase.sdf")
#logging.getLogger("foqus." + __name__).info(
# "TurbineLite Database:\n {0}".format(db.dbFile))
#add 'foqus' app to TurbineLite DB if not already there
#db.add_new_application('foqus')
#register the consumer in the database
db.consumer_register()
#print("consumer_uuid: {0}".format(consumer_uuid))
#write the time to the turbineLite db about every minute
kat = _KeepAliveTimer(db, freq=60)
kat.start()
guid = job_desc['Id']
jid = None
simId = job_desc['Simulation']
# Run the job
db.add_message("consumer={0}, starting job {1}"\
.format(db.consumer_id, jid), guid)
db.job_change_status(guid, "setup")
configContent = db.get_configuration_file(simId)
logging.getLogger("foqus." + __name__)\
.info("Job {0} is submitted".format(jid))
#db.jobConsumerID(guid, consumer_uuid)
db.job_prepare(guid, jid, configContent)
# Load the session file
dat.load(sfile, stopConsumers=True)
dat.loadFlowsheetValues(vfile)
'''
"flowsheet"
"nodes": {
"BFB": {
"browser_conf": null,
"modelName": "BFB_v11_FBS_01_26_2018",
"pythonCode": "#run steady state init\nself.options[\"Script\"].value = \"Run_Steady\"\nself.runModel()\nif self.calcError != -1:\n raise(Exception(\"Steady state homotopy failed\"))\n#Run optimization\nself.options[\"Script\"].value = \"Init_Opt\"\nself.runModel()\nif self.calcError != -1:\n raise(Exception(\"Optimization failed\"))\n# Update the x and f dicts from the node output\n# f gets copied to the node outputs so need this \n# for now x doesn't get copied back\nx, f = self.getValues()",
"calcError": 0,
"turbSession": "9c9dff4f-48b9-482a-99be-1bfe879350f5",
"dmf_sim_ids": null,
"scriptMode": "total",
"turbApp": [
"ACM",
"aspenfile"
],
"synced": true,
"modelType": 2,
"y": 0,
"x": -200,
"z": 0,
"options": {...}
'''
# dat.flowsheet.nodes.
s3 = boto3.client('s3', region_name='us-east-1')
bucket_name = 'foqus-simulations'
l = s3.list_objects(Bucket=bucket_name,
Prefix='anonymous/%s' % job_desc['Simulation'])
if not l.has_key('Contents'):
_log.info("S3 Simulation: No keys match %s" % 'anonymous/%s' %
job_desc['Simulation'])
return
_log.debug("FLOWSHEET NODES")
for nkey in dat.flowsheet.nodes:
if dat.flowsheet.nodes[nkey].turbApp is None:
continue
assert len(dat.flowsheet.nodes[nkey].turbApp) == 2, \
'DAT Flowsheet nodes turbApp is %s' %dat.flowsheet.nodes[nkey].turbApp
node = dat.flowsheet.nodes[nkey]
model_name = node.modelName
sinter_filename = 'anonymous/%s/%s/%s.json' % (job_desc['Simulation'],
nkey, model_name)
s3_key_list = map(lambda i: i['Key'], l['Contents'])
assert sinter_filename in s3_key_list, 'missing sinter configuration "%s" not in %s' % (
sinter_filename, str(s3_key_list))
simulation_name = job_desc.get('Simulation')
#sim_list = node.gr.turbConfig.getSimulationList()
sim_list = turbine_simulation_script.main_list(
[node.gr.turbConfig.getFile()])
print("===" * 20)
print(sim_list)
print("===" * 20)
sim_d = filter(lambda i: i['Name'] == model_name, sim_list)
assert len(
sim_d
) < 2, 'Expecting 0 or 1 entries for simulation %s' % simulation_name
if len(sim_d) == 0:
sim_d = None
else:
sim_d = sim_d[0]
if dat.flowsheet.nodes[nkey].turbApp[0] == 'ACM':
model_filename = 'anonymous/%s/%s/%s.acmf' % (simulation_name,
nkey, model_name)
assert model_filename in s3_key_list, 'missing sinter configuration "%s"' % sinter_filename
else:
raise NotImplementedError, 'Flowsheet Node model type: "%s"' % (
str(dat.flowsheet.nodes[nkey].turbApp))
prefix = 'anonymous/%s/%s/' % (job_desc['Simulation'], nkey)
entry_list = filter(
lambda i: i['Key'] != prefix and i['Key'].startswith(prefix),
l['Contents'])
sinter_local_filename = None
update_required = False
for entry in entry_list:
_log.debug("ENTRY: %s", entry)
key = entry['Key']
etag = entry.get('ETag', "").strip('"')
file_name = key.split('/')[-1]
file_path = os.path.join(workingDirectory, file_name)
# Upload to TurbineLite
# if ends with json or acmf
si_metadata = []
if key.endswith('.json'):
_log.debug('CONFIGURATION FILE')
sinter_local_filename = file_path
if sim_d:
si_metadata = filter(
lambda i: i['Name'] == 'configuration',
sim_d["StagedInputs"])
elif key.endswith('.acmf'):
_log.debug('ACMF FILE')
if sim_d:
si_metadata = filter(lambda i: i['Name'] == 'aspenfile',
sim_d["StagedInputs"])
else:
raise NotImplementedError, 'Not allowing File "%s" to be staged in' % key
assert len(
si_metadata
) < 2, 'Turbine Error: Too many entries for "%s", "%s"' % (
simulation_name, file_name)
# NOTE: Multipart uploads have different ETags ( end with -2 or something )
# THus the has comparison will fail
# FOr now ignore it, but fixing this check is performance optimization.
#
if len(si_metadata) == 1:
file_hash = si_metadata[0]['MD5Sum']
if file_hash.lower() != etag.lower():
_log.debug("Compare %s: %s != %s" %
(file_name, etag, file_hash))
_log.debug('s3 download(%s): %s' % (workingDirectory, key))
s3.download_file(bucket_name, key, file_path)
update_required = True
else:
_log.debug("MATCH")
s3.download_file(bucket_name, key, file_path)
else:
_log.debug("Add to Turbine Simulation(%s) File: %s" %
(simulation_name, file_name))
s3.download_file(bucket_name, key, file_path)
update_required = True
assert sinter_local_filename is not None, 'missing sinter configuration file'
if model_name not in map(lambda i: i['Name'], sim_list):
_log.debug('Adding Simulation "%s"' % model_name)
node.gr.turbConfig.uploadSimulation(model_name,
sinter_local_filename,
update=False)
elif update_required:
# NOTE: Requires the configuration file on update, so must download_file it above...
_log.debug('Updating Simulation "%s"' % model_name)
node.gr.turbConfig.uploadSimulation(model_name,
sinter_local_filename,
update=True)
else:
_log.debug('No Update Required for Simulation "%s"' % model_name)
db.job_change_status(guid, "running")
gt = dat.flowsheet.runAsThread()
terminate = False
while gt.isAlive():
gt.join(10)
status = db.consumer_status()
if status == 'terminate':
terminate = True
db.job_change_status(guid, "error")
gt.terminate()
break
if terminate:
return
if gt.res[0]:
dat.flowsheet.loadValues(gt.res[0])
else:
dat.flowsheet.errorStat = 19
dat.saveFlowsheetValues(ofile)
db.job_save_output(guid, workingDirectory)
dat.save(filename=rfile,
updateCurrentFile=False,
changeLogMsg="Saved Turbine Run",
bkp=False,
indent=0)
if dat.flowsheet.errorStat == 0:
db.job_change_status(guid, "success")
db.add_message(
"consumer={0}, job {1} finished, success"\
.format(db.consumer_id, jid), guid)
else:
db.job_change_status(guid, "error")
db.add_message(
"consumer={0}, job {1} finished, error"\
.format(db.consumer_id, jid), guid)
logging.getLogger("foqus." + __name__)\
.info("Job {0} finished"\
.format(jid))
#stop all Turbine consumers
dat.flowsheet.turbConfig.stopAllConsumers()
dat.flowsheet.turbConfig.closeTurbineLiteDB()
sys.exit(exit_code)
def test_GET_List(self):
l = tss.main_list([self.config_name], func=None)
self.log.debug('simulation names %s' %
map(lambda f: (f['Name'], f['Id']), l))
self.assertTrue(type(l) is list, 'return type should be list')