class NewtIntegrationTest(BaseIntegrationTest):
def __init__(self, name, girder_url, girder_user, girder_password, machine, job_timeout=60 * 5):
super(NewtIntegrationTest, self).__init__(name, girder_url, girder_user, girder_password, job_timeout)
self._cluster_id = None
self._machine = machine
def setUp(self):
# First authenticate with NEWT
self._session = Session()
r = self._session.post(
"https://newt.nersc.gov/newt/auth", {"username": self._girder_user, "password": self._girder_password}
)
self.assertEqual(r.status_code, 200)
print r.json()
self._newt_session_id = r.json()["newt_sessionid"]
# Now authenticate with Girder using the session id
url = "%s/api/v1/newt/authenticate/%s" % (self._girder_url, self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = "%s/api/v1/newt/authenticate/%s" % (self._girder_url, self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = "%s/api/v1" % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.token = self._session.cookies["girderToken"]
user = self._client.get("user/me")
self._user_id = user["_id"]
r = self._client.listFolder(self._user_id, "user", name="Private")
r = list(r)
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]["_id"]
def tearDown(self):
super(NewtIntegrationTest, self).tearDown()
if self._cluster_id:
try:
url = "clusters/%s" % self._cluster_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_cluster(self):
body = {"config": {"host": self._machine}, "name": "NewtIntegrationTest", "type": "newt"}
r = self._client.post("clusters", data=json.dumps(body))
self._cluster_id = r["_id"]
# Now test the connection
r = self._client.put("clusters/%s/start" % self._cluster_id)
sleeps = 0
while True:
time.sleep(1)
r = self._client.get("clusters/%s/status" % self._cluster_id)
if r["status"] == "running":
break
elif r["status"] == "error":
r = self._client.get("clusters/%s/log" % self._cluster_id)
self.fail(str(r))
if sleeps > 9:
self.fail("Cluster never moved into running state")
sleeps += 1
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i["name"].startswith("CumulusIntegrationTestJob-%s.o" % self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get("item/%s/files" % i["_id"])
self.assertEqual(len(r), 1)
url = "%s/api/v1/file/%s/download" % (self._girder_url, r[0]["_id"])
r = self._session.get(url)
self.assertEqual(r.content, self._data)
def test(self):
try:
self.create_cluster()
self.create_script()
self.create_input()
self.create_output_folder()
self.create_job()
self.submit_job(timeout=self._job_timeout)
self.assert_output()
except HttpError as error:
self.fail(error.responseText)
def main(config):
client = GirderClient(apiUrl=config.girder_api_url)
client.authenticate(config.girder_user,
config.girder_password)
# Load any parameters
params = {}
if config.taskflow_start_params is not None:
with open(config.taskflow_start_params) as fp:
params = json.load(fp)
print params
try:
print ('Running %s taskflow ...' % config.taskflow_start_params)
taskflow_id = create_taskflow(
client, config.taskflow_class)
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url, data=json.dumps(params))
# Wait for it to complete
wait_for_complete(client, taskflow_id)
except HttpError as ex:
print( ex.responseText)
def _taskflow_task_finished(taskflow, taskflow_task_id):
girder_token = taskflow['girder_token']
girder_api_url = taskflow['girder_api_url']
client = GirderClient(apiUrl=girder_api_url)
client.token = girder_token
url = 'taskflows/%s/tasks/%s/finished' % (taskflow.id, taskflow_task_id)
return client.put(url)
def _taskflow_task_finished(taskflow, taskflow_task_id):
girder_token = taskflow['girder_token']
girder_api_url = taskflow['girder_api_url']
client = GirderClient(apiUrl=girder_api_url)
client.token = girder_token
url = 'taskflows/%s/tasks/%s/finished' % (taskflow.id, taskflow_task_id)
return client.put(url)
lastName='Admin')
client.authenticate(user, password)
s3_assetstore_name = 's3'
if find_assetstore(s3_assetstore_name) is None:
client.post('assetstore',
parameters=dict(name=s3_assetstore_name,
type=str(AssetstoreType.S3),
bucket=args.s3,
accessKeyId=args.aws_key_id,
secret=args.aws_secret_key))
client.put(
'system/plugins',
parameters=dict(plugins=json.dumps(['jobs', 'worker', 'osumo']))
)
client.put('system/restart')
sleep(30)
client.put('system/setting',
parameters=dict(list=json.dumps([
dict(key='worker.broker', value=args.broker),
dict(key='worker.backend', value=args.broker),
dict(key='core.route_table', value=dict(
core_girder="/girder",
core_static_root="/static",
osumo="/"))])))
client.put('system/restart')
class CumulusClient():
'''Application interface to cumulus-based client for HPC systems
supporting NEWT API.
Note: the methods must be called in a specific order!
create_cluster()
create_omega3p_script()
create_job()
upload_inputs()
submit_job()
Then optionally:
monitor_job()
download_results()
release_resources()
'''
# ---------------------------------------------------------------------
def __init__(self, girder_url, newt_sessionid):
'''
'''
self._client = None
self._cluster_id = None
self._girder_url = girder_url
self._input_folder_id = None
self._job_folder_id = None
self._job_id = None
self._output_folder_id = None
self._private_folder_id = None
self._script_id = None
self._session = requests.Session()
# Authenticate with Girder using the newt session id
url = '%s/api/v1/newt/authenticate/%s' % \
(self._girder_url, newt_sessionid)
r = self._session.put(url)
if r.status_code != 200:
raise HttpError(r.status_code, r.text, r.url, r.request.method)
# Instantiate Girder client
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.token = self._session.cookies['girderToken']
user = self._client.get('user/me')
#print 'user', user
user_id = user['_id']
r = self._client.listFolder(user_id, 'user', name='Private')
if len(r) != 1:
raise Exception('Wrong number of users; should be 1 got %s' %
len(r))
self._private_folder_id = r[0]['_id']
print 'private_folder_id', self._private_folder_id
# ---------------------------------------------------------------------
def job_id(self):
'''Returns current job id (which may be None)
'''
return self._job_id
# ---------------------------------------------------------------------
def create_cluster(self, machine_name, cluster_name=None):
'''
'''
if cluster_name is None:
user = self._client.get('user/me')
user_name = user.get('firstName', 'user')
cluster_name = '%s.%s' % (machine_name, user_name)
cluster = None
cluster_list = self._client.get('clusters')
for extant_cluster in cluster_list:
if extant_cluster['name'] == cluster_name:
cluster = extant_cluster
self._cluster_id = extant_cluster['_id']
break
if not cluster:
body = {
'config': {
'host': machine_name
},
'name': cluster_name,
'type': 'newt'
}
r = self._client.post('clusters', data=json.dumps(body))
self._cluster_id = r['_id']
print 'cluster_id', self._cluster_id
# Reset the state of the cluster
body = {'status': 'created'}
r = self._client.patch('clusters/%s' % self._cluster_id,
data=json.dumps(body))
# Now test the connection
r = self._client.put('clusters/%s/start' % self._cluster_id)
sleeps = 0
while True:
time.sleep(1)
r = self._client.get('clusters/%s/status' % self._cluster_id)
if r['status'] == 'running':
break
elif r['status'] == 'error':
r = self._client.get('clusters/%s/log' % self._cluster_id)
print r
raise Exception('ERROR creating cluster')
if sleeps > 9:
raise Exception('Cluster never moved into running state')
sleeps += 1
# ---------------------------------------------------------------------
def create_omega3p_script(self,
omega3p_filename,
name=None,
number_of_tasks=1):
'''Creates script to submit omega3p job
'''
command = 'srun -n %s /project/projectdirs/ace3p/{{machine}}/omega3p %s' % \
(number_of_tasks, omega3p_filename)
if name is None:
name = omega3p_filename
body = {'commands': [command], 'name': name}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
print 'script_id', self._script_id
# ---------------------------------------------------------------------
def create_input(self, input_paths, folder_name='input_files'):
'''DEPRECATED Uploads input files
'''
folder_id = self.get_folder(self._private_folder_id, folder_name)
if folder_id is None:
return
print 'input_folder_id', folder_id
self._input_folder_id = folder_id
def upload_file(path):
name = os.path.basename(path)
size = os.path.getsize(path)
with open(path, 'rb') as fp:
self._client.uploadFile(self._input_folder_id,
fp,
name,
size,
parentType='folder')
for input_path in input_paths:
if not input_path or not os.path.exists(input_path):
raise Exception('Input file not found: %s' % input_path)
upload_file(input_path)
# ---------------------------------------------------------------------
def create_output_folder(self, folder_name='output_files'):
'''DEPRECATED
'''
folder_id = self.get_folder(self._private_folder_id, folder_name)
print 'output_folder_id', folder_id
self._output_folder_id = folder_id
# ---------------------------------------------------------------------
def create_job(self, job_name, tail=None):
'''
'''
# Create job folders
folder_name = uuid.uuid4().hex # unique name
self._job_folder_id = self.get_folder(self._private_folder_id,
folder_name)
print 'Created job folder', folder_name
self._input_folder_id = self.get_folder(self._job_folder_id,
'input_files')
self._output_folder_id = self.get_folder(self._job_folder_id,
'output_files')
# Make sure job_name isn't null
if not job_name:
job_name = 'CumulusJob'
# Create job spec
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [{
'folderId': self._input_folder_id,
'path': '.'
}]
}
if tail:
body['output'].append({"path": tail, "tail": True})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
print 'Created job_id', self._job_id
# ---------------------------------------------------------------------
def upload_inputs(self, input_paths):
'''Uploads input files to input folder
'''
if not self._input_folder_id:
raise Exception('Input folder missing')
def upload_file(path):
name = os.path.basename(path)
size = os.path.getsize(path)
with open(path, 'rb') as fp:
self._client.uploadFile(self._input_folder_id,
fp,
name,
size,
parentType='folder')
for input_path in input_paths:
if not input_path or not os.path.exists(input_path):
raise Exception('Input file not found: %s' % input_path)
upload_file(input_path)
# ---------------------------------------------------------------------
def submit_job(self,
machine,
project_account,
timeout_minutes,
queue='debug',
qos=None,
number_of_nodes=1,
job_output_dir=None):
'''
'''
body = {
'machine': machine,
'account': project_account,
'numberOfNodes': number_of_nodes,
'maxWallTime': {
'hours': 0,
'minutes': timeout_minutes,
'seconds': 0
},
'queue': queue,
}
if qos:
body['qualityOfService'] = qos
#print 'jobOutputDir', job_output_dir
if job_output_dir:
body['jobOutputDir'] = job_output_dir
print 'Setting jobOutputDir', job_output_dir
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(body))
print 'Submitted job', self._job_id
# ---------------------------------------------------------------------
def monitor_job(self, tail=None):
'''Periodically monitors job status
'''
log_offset = 0
job_timeout = 60 * timeout_minutes
start = time.time()
while True:
time.sleep(2)
# Provide some feedback at startup
if log_offset == 0:
sys.stdout.write('.')
#print 'Checking status'
r = self._client.get('jobs/%s' % self._job_id)
#print r
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
raise Exception(str(r))
elif r['status'] == 'complete':
break
# Tail log file
if tail:
params = {'offset': log_offset, 'path': tail}
#print 'Checking tail'
r = self._client.get('jobs/%s/output' % self._job_id,
parameters=params)
#print r
output = r['content']
if output and log_offset == 0:
print # end the user feedback dots
log_offset += len(output)
for l in output:
print l
sys.stdout.flush()
if time.time() - start > job_timeout:
raise Exception('Job timeout')
# ---------------------------------------------------------------------
def download_results(self, destination_folder):
'''Downloads all output files to a local directory
'''
if not os.path.exists(destination_folder):
os.makedirs(destination_folder)
self._client.downloadFolderRecursive(self._output_folder_id,
destination_folder)
print 'Downloaded files to %s' % destination_folder
# ---------------------------------------------------------------------
def release_resources(self):
'''Closes/deletes any current resources
'''
resource_info = {
'clusters': [self._cluster_id],
'jobs': [self._job_id],
'scripts': [self._script_id],
'folder': [self._job_folder]
}
for resource_type, id_list in resource_info.items():
for resource_id in id_list:
if resource_id is not None:
url = '%s/%s' % (resource_type, resource_id)
self._client.delete(url)
self._input_folder_id = None
self._job_folder_id = None
self._job_id = None
self._output_folder_id = None
self._script_id = None
# ---------------------------------------------------------------------
def get_folder(self, parent_id, name):
'''Returns folder_id, creating one if needed
'''
# Check if folder already exists
folder_list = self._client.listFolder(parent_id, name=name)
if folder_list:
folder = folder_list[0]
#print 'found folder %s: %s' % (name, str(folder))
return folder['_id']
# (else)
try:
r = self._client.createFolder(parent_id, name)
return r['_id']
except HttpError as e:
print e.responseText
return None
args = parser.parse_args()
# Get the ID for our Analyses folder.
c = GirderClient(host=args.girder_host, port=args.girder_port)
c.authenticate('girder', 'girder')
folderSearch = c.get('resource/search', parameters={
'q': 'Analyses',
'types': '["folder"]'
})
folderId = folderSearch['folder'][0]['_id']
# Disable authorization requirements for running romanesco tasks
c.put('system/setting', parameters={
'key': 'flow.require_auth',
'value': 'false'
})
# Check if these analyses already exist. If so, we won't re-upload them.
uploadACR = False
uploadPGS = False
searchACR = c.get('resource/search', {
'q': 'aceArbor',
'types': '["item"]'
})
if len(searchACR['item']) == 0:
uploadACR = True
searchPGS = c.get('resource/search', {
'q': 'Phylogenetic signal',
class NewtIntegrationTest(BaseIntegrationTest):
def __init__(self,
name,
girder_url,
girder_user,
girder_password,
machine,
job_timeout=60 * 5):
super(NewtIntegrationTest,
self).__init__(name, girder_url, girder_user, girder_password,
job_timeout)
self._cluster_id = None
self._machine = machine
def setUp(self):
# First authenticate with NEWT
self._session = Session()
r = self._session.post('https://newt.nersc.gov/newt/auth', {
'username': self._girder_user,
'password': self._girder_password
})
self.assertEqual(r.status_code, 200)
print r.json()
self._newt_session_id = r.json()['newt_sessionid']
# Now authenticate with Girder using the session id
url = '%s/api/v1/newt/authenticate/%s' % (self._girder_url,
self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = '%s/api/v1/newt/authenticate/%s' % (self._girder_url,
self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.token = self._session.cookies['girderToken']
user = self._client.get('user/me')
self._user_id = user['_id']
r = self._client.listFolder(self._user_id, 'user', name='Private')
r = list(r)
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
super(NewtIntegrationTest, self).tearDown()
if self._cluster_id:
try:
url = 'clusters/%s' % self._cluster_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_cluster(self):
body = {
'config': {
'host': self._machine
},
'name': 'NewtIntegrationTest',
'type': 'newt'
}
r = self._client.post('clusters', data=json.dumps(body))
self._cluster_id = r['_id']
# Now test the connection
r = self._client.put('clusters/%s/start' % self._cluster_id)
sleeps = 0
while True:
time.sleep(1)
r = self._client.get('clusters/%s/status' % self._cluster_id)
if r['status'] == 'running':
break
elif r['status'] == 'error':
r = self._client.get('clusters/%s/log' % self._cluster_id)
self.fail(str(r))
if sleeps > 9:
self.fail('Cluster never moved into running state')
sleeps += 1
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' %
self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
url = '%s/api/v1/file/%s/download' % (self._girder_url, r[0]['_id'])
r = self._session.get(url)
self.assertEqual(r.content, self._data)
def test(self):
try:
self.create_cluster()
self.create_script()
self.create_input()
self.create_output_folder()
self.create_job()
self.submit_job(timeout=self._job_timeout)
self.assert_output()
except HttpError as error:
self.fail(error.responseText)
class NewtIntegrationTest(BaseIntegrationTest):
def __init__(self, name, girder_url, girder_user, girder_password, machine,
job_timeout=60*5):
super(NewtIntegrationTest, self).__init__(name, girder_url, girder_user,
girder_password, job_timeout)
self._cluster_id = None
self._machine = machine
def setUp(self):
# First authenticate with NEWT
self._session = Session()
r = self._session.post('https://newt.nersc.gov/newt/auth',
{
'username': self._girder_user,
'password': self._girder_password})
self.assertEqual(r.status_code, 200)
print r.json()
self._newt_session_id = r.json()['newt_sessionid']
# Now authenticate with Girder using the session id
url = '%s/api/v1/newt/authenticate/%s' % (self._girder_url, self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = '%s/api/v1/newt/authenticate/%s' % (self._girder_url, self._newt_session_id)
r = self._session.put(url)
self.assertEqual(r.status_code, 200)
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.token = self._session.cookies['girderToken']
user = self._client.get('user/me')
self._user_id = user['_id']
r = self._client.listFolder(self._user_id, 'user', name='Private')
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
super(NewtIntegrationTest, self).tearDown()
if self._cluster_id:
try:
url = 'clusters/%s' % self._cluster_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_cluster(self):
body = {
'config': {
'host': self._machine
},
'name': 'NewtIntegrationTest',
'type': 'newt'
}
r = self._client.post('clusters', data=json.dumps(body))
self._cluster_id = r['_id']
# Now test the connection
r = self._client.put('clusters/%s/start' % self._cluster_id)
sleeps = 0
while True:
time.sleep(1)
r = self._client.get('clusters/%s/status' % self._cluster_id)
if r['status'] == 'running':
break
elif r['status'] == 'error':
r = self._client.get('clusters/%s/log' % self._cluster_id)
self.fail(str(r))
if sleeps > 9:
self.fail('Cluster never moved into running state')
sleeps += 1
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' % self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
url = '%s/api/v1/file/%s/download' % (self._girder_url, r[0]['_id'])
r = self._session.get(url)
self.assertEqual(r.content, self._data)
def test(self):
try:
self.create_cluster()
self.create_script()
self.create_input()
self.create_output_folder()
self.create_job()
self.submit_job(timeout=self._job_timeout)
self.assert_output()
except HttpError as error:
self.fail(error.responseText)
if __name__ == "__main__":
login = '******'
password = '******'
gc = GirderClient(apiUrl='http://*****:*****@admin.com',
firstName='admin',
lastName='admin',
password=password,
admin=True)
gc.authenticate(username=login, password=password)
# Create an assetstore
gc.post('assetstore',
parameters={
'name': 'TestAssetstore',
'type': 0,
'root': '/home/circleci/project/assetstore'
})
# Enable the 'slicer_extension_manager' plugin
gc.put('system/plugins',
parameters={"plugins": '["slicer_extension_manager"]'})
# Restart the server
gc.put('system/restart')
class BaseIntegrationTest(unittest.TestCase):
def __init__(self, name, girder_url, girder_user, girder_password, job_timeout=60, cleanup=True):
super(BaseIntegrationTest, self).__init__(name)
self._job_id = None
self._script_id = None
self._output_folder_id = None
self._input_folder_id = None
self._girder_url = girder_url
self._girder_user = girder_user
self._girder_password = girder_password
self._job_timeout = job_timeout
self._data = 'Need more input!'
self._cleanup = cleanup
def setUp(self):
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.authenticate(self._girder_user,
self._girder_password)
user = self._client.get('user/me')
self._user_id = user['_id']
r = list(self._client.listFolder(self._user_id, 'user', name='Private'))
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
if not self._cleanup:
return
if self._job_id:
try:
url = 'jobs/%s' % self._job_id
self._client.delete(url)
except Exception as e:
traceback.print_exc()
if self._script_id:
try:
url = 'scripts/%s' % self._script_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._output_folder_id:
try:
url = 'folder/%s' % self._output_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._input_folder_id:
try:
url = 'folder/%s' % self._input_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_script(self, commands=[
'sleep 10', 'cat CumulusIntegrationTestInput'
]):
body = {
'commands': commands,
'name': 'CumulusIntegrationTestLob'
}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
def create_input(self, folder_name='CumulusInput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._input_folder_id = r['_id']
size = len(self._data)
item = self._client.uploadFile(self._input_folder_id,
StringIO(self._data), 'CumulusIntegrationTestInput', size,
parentType='folder')
self._item_id = item['itemId']
def create_output_folder(self, folder_name='CumulusOutput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._output_folder_id = r['_id']
def create_job(self, job_name='CumulusIntegrationTestJob', tail=None):
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [
{
'folderId': self._input_folder_id,
'path': '.'
}
]
}
if tail:
body['output'].append({
"path": tail,
"tail": True
})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
def submit_job(self, job_params={}, timeout=None):
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(job_params))
start = time.time()
while True:
time.sleep(1)
r = self._client.get('jobs/%s' % self._job_id)
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
self.fail(str(r))
elif r['status'] == 'complete':
break
if time.time() - start > timeout:
self.fail('Job didn\'t complete in timeout')
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' % self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
path = os.path.join(tempfile.gettempdir(), self._job_id)
try:
self._client.downloadFile(r[0]['_id'], path)
with open(path, 'rb') as fp:
self.assertEqual(fp.read(), self._data)
finally:
if os.path.exists(path):
os.remove(path)
local_assetstore_name = 'local'
if find_assetstore(local_assetstore_name) is None:
client.post('assetstore',
parameters=dict(name=local_assetstore_name,
type=str(AssetstoreType.FILESYSTEM),
root=os.path.join(args.data_root,
'assetstores',
'local'),
readOnly="false"))
client.put('system/plugins',
parameters=dict(plugins=json.dumps(['celery_jobs',
'climos_test',
'user_quota',
'hdfs_assetstore',
'jobs',
'romanesco',
'sparktest'])))
client.put('system/restart')
sleep(30)
client.put('system/setting',
parameters=dict(list=json.dumps([
dict(key='romanesco.require_auth', value=False),
dict(key='romanesco.broker', value=args.broker),
dict(key='romanesco.backend', value=args.broker)])))
hdfs_assetstore_name = 'hdfs'
client.post('user',
parameters=dict(login=username,
password=password,
email=kwds['email'],
firstName=kwds['firstName'],
lastName=kwds['lastName']))
client = GirderClient(host='localhost', port=8080)
if find_user('girder'):
client.authenticate('girder', 'girder')
ensure_user(client,
login='******',
password='******',
email='*****@*****.**',
firstName='girder',
lastName='girder')
client.authenticate('girder', 'girder')
client.put(
'system/setting',
parameters=dict(list=json.dumps([
dict(key='worker.broker', value='amqp://guest@localhost'),
dict(key='worker.backend', value='amqp://guest@localhost'),
dict(key='core.route_table',
value=dict(
core_girder='/girder', core_static_root='/static', osumo='/'))
])))
'public': 'true'
})
c.post('folder',
parameters={
'parentType': 'collection',
'parentId': collection['_id'],
'name': 'Data',
'description': 'Data Folder',
'public': 'true'
})
c.post('folder',
parameters={
'parentType': 'collection',
'parentId': collection['_id'],
'name': 'Analyses',
'description': 'Analysis folder',
'public': 'true'
})
# Turn on the romanesco plugin
c.put('system/plugins', parameters={"plugins": '["romanesco"]'})
# Create an assetstore if there isn't one
if len(c.get('assetstore')) == 0:
c.post('assetstore',
parameters={
'type': '1',
'name': 'GridFS',
'db': 'girder-gridfs'
})
collection = c.post(
"collection", parameters={"name": "Default", "description": "Default workspace", "public": "true"}
)
c.post(
"folder",
parameters={
"parentType": "collection",
"parentId": collection["_id"],
"name": "Data",
"description": "Data Folder",
"public": "true",
},
)
c.post(
"folder",
parameters={
"parentType": "collection",
"parentId": collection["_id"],
"name": "Analyses",
"description": "Analysis folder",
"public": "true",
},
)
# Turn on the romanesco plugin
c.put("system/plugins", parameters={"plugins": '["romanesco"]'})
# Create an assetstore if there isn't one
if len(c.get("assetstore")) == 0:
c.post("assetstore", parameters={"type": "1", "name": "GridFS", "db": "girder-gridfs"})
def main(config):
client = GirderClient(apiUrl=config.girder_api_url)
client.authenticate(config.girder_user,
config.girder_password)
try:
# First run the simple flow
print ('Running simple taskflow ...')
taskflow_id = create_taskflow(
client, 'cumulus.taskflow.core.test.mytaskflows.SimpleTaskFlow')
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url)
# Wait for it to complete
wait_for_complete(client, taskflow_id)
# First run the simple flow
print ('Running linked taskflow ...')
taskflow_id = create_taskflow(
client, 'cumulus.taskflow.core.test.mytaskflows.LinkTaskFlow')
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url)
# Wait for it to complete
wait_for_complete(client, taskflow_id)
# Test terminating a simple flow
print ('Running simple taskflow ...')
taskflow_id = create_taskflow(
client, 'cumulus.taskflow.core.test.mytaskflows.SimpleTaskFlow')
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url)
time.sleep(4)
print ('Terminate the taskflow')
url = 'taskflows/%s/terminate' % (taskflow_id)
client.put(url)
# Wait for it to terminate
wait_for_terminated(client, taskflow_id)
# Now delete it
print ('Delete the taskflow')
url = 'taskflows/%s' % (taskflow_id)
try:
client.delete(url)
except HttpError as ex:
if ex.status != 202:
raise
# Wait for it to terminate
wait_for_deletion(client, taskflow_id)
# Now try something with a chord
print ('Running taskflow containing a chord ...')
taskflow_id = create_taskflow(
client, 'cumulus.taskflow.core.test.mytaskflows.ChordTaskFlow')
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url)
# Wait for it to complete
wait_for_complete(client, taskflow_id)
# Now try a workflow that is the two connected together
print ('Running taskflow that connects to parts together ...')
taskflow_id = create_taskflow(
client, 'cumulus.taskflow.core.test.mytaskflows.ConnectTwoTaskFlow')
# Start the task flow
url = 'taskflows/%s/start' % (taskflow_id)
client.put(url)
# Wait for it to complete
wait_for_complete(client, taskflow_id)
# # Now try a composite workflow approach ...
# print ('Running taskflow that is a composite ...')
# taskflow_id = create_taskflow(
# client, 'cumulus.taskflow.core.test.mytaskflows.MyCompositeTaskFlow')
#
# # Start the task flow
# url = 'taskflows/%s/start' % (taskflow_id)
# client.put(url)
#
# # Wait for it to complete
# wait_for_complete(client, taskflow_id)
except HttpError as ex:
print( ex.responseText)
firstName='Girder',
lastName='Admin')
client.authenticate(user, password)
s3_assetstore_name = 's3'
if find_assetstore(s3_assetstore_name) is None:
client.post('assetstore',
parameters=dict(name=s3_assetstore_name,
type=str(AssetstoreType.S3),
bucket=args.s3,
accessKeyId=args.aws_key_id,
secret=args.aws_secret_key))
client.put('system/plugins',
parameters=dict(plugins=json.dumps(['jobs', 'worker', 'osumo'])))
client.put('system/restart')
sleep(30)
client.put(
'system/setting',
parameters=dict(list=json.dumps([
dict(key='worker.broker', value=args.broker),
dict(key='worker.backend', value=args.broker),
dict(key='core.route_table',
value=dict(
core_girder="/girder", core_static_root="/static", osumo="/"))
])))
client.put('system/restart')
offset += limit
return result
client = GirderClient(host='localhost', port=8080)
if find_user('girder'):
client.authenticate('girder', 'girder')
ensure_user(client,
login='******',
password='******',
email='*****@*****.**',
firstName='girder',
lastName='girder')
client.authenticate('girder', 'girder')
if find_assetstore('local') is None:
client.post('assetstore',
parameters=dict(name='local',
type=str(AssetstoreType.GRIDFS),
db='sumoLocalStore',
mongohost='mongodb://localhost:27017',
replicaset=''))
client.put('system/plugins',
parameters=dict(plugins=json.dumps(['jobs', 'worker', 'osumo'])))
arborWebAppsPath = sys.argv[1]
# Get the ID for our Analyses folder.
c = GirderClient(host='localhost', port=9000)
c.authenticate('girder', 'girder')
folderSearch = c.get('resource/search',
parameters={
'q': 'Analyses',
'types': '["folder"]'
})
folderId = folderSearch['folder'][0]['_id']
# Disable authorization requirements for running romanesco tasks
c.put('system/setting',
parameters={
'key': 'romanesco.require_auth',
'value': 'false'
})
# Check if these analyses already exist. If so, we won't re-upload them.
uploadACR = False
uploadPGS = False
searchACR = c.get('resource/search', {'q': 'aceArbor', 'types': '["item"]'})
if len(searchACR['item']) == 0:
uploadACR = True
searchPGS = c.get('resource/search', {
'q': 'Phylogenetic signal',
'types': '["item"]'
})
class BaseIntegrationTest(unittest.TestCase):
def __init__(self,
name,
girder_url,
girder_user,
girder_password,
job_timeout=60,
cleanup=True):
super(BaseIntegrationTest, self).__init__(name)
self._job_id = None
self._script_id = None
self._output_folder_id = None
self._input_folder_id = None
self._girder_url = girder_url
self._girder_user = girder_user
self._girder_password = girder_password
self._job_timeout = job_timeout
self._data = 'Need more input!'
self._cleanup = cleanup
def setUp(self):
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.authenticate(self._girder_user, self._girder_password)
user = self._client.get('user/me')
self._user_id = user['_id']
r = list(self._client.listFolder(self._user_id, 'user',
name='Private'))
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
if not self._cleanup:
return
if self._job_id:
try:
url = 'jobs/%s' % self._job_id
self._client.delete(url)
except Exception as e:
traceback.print_exc()
if self._script_id:
try:
url = 'scripts/%s' % self._script_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._output_folder_id:
try:
url = 'folder/%s' % self._output_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._input_folder_id:
try:
url = 'folder/%s' % self._input_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_script(self,
commands=['sleep 10',
'cat CumulusIntegrationTestInput']):
body = {'commands': commands, 'name': 'CumulusIntegrationTestLob'}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
def create_input(self, folder_name='CumulusInput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._input_folder_id = r['_id']
size = len(self._data)
item = self._client.uploadFile(self._input_folder_id,
StringIO(self._data),
'CumulusIntegrationTestInput',
size,
parentType='folder')
self._item_id = item['itemId']
def create_output_folder(self, folder_name='CumulusOutput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._output_folder_id = r['_id']
def create_job(self, job_name='CumulusIntegrationTestJob', tail=None):
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [{
'folderId': self._input_folder_id,
'path': '.'
}]
}
if tail:
body['output'].append({"path": tail, "tail": True})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
def submit_job(self, job_params={}, timeout=None):
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(job_params))
start = time.time()
while True:
time.sleep(1)
r = self._client.get('jobs/%s' % self._job_id)
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
self.fail(str(r))
elif r['status'] == 'complete':
break
if time.time() - start > timeout:
self.fail('Job didn\'t complete in timeout')
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' %
self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
path = os.path.join(tempfile.gettempdir(), self._job_id)
try:
self._client.downloadFile(r[0]['_id'], path)
with open(path, 'rb') as fp:
self.assertEqual(fp.read(), self._data)
finally:
if os.path.exists(path):
os.remove(path)