def armada_base(self, context):
# Define task_instance
self.task_instance = context['task_instance']
# Set up and retrieve values from xcom
self.xcom_puller = XcomPuller(self.main_dag_name, self.task_instance)
self.action_info = self.xcom_puller.get_action_info()
self.dc = self.xcom_puller.get_deployment_configuration()
# Set up xcom_pusher to push values to xcom
self.xcom_pusher = XcomPusher(self.task_instance)
# Logs uuid of action performed by the Operator
logging.info("Armada Operator for action %s", self.action_info['id'])
# Retrieve Endpoint Information
armada_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.armada_svc_type)
# Set up armada client
self.armada_client = self._init_armada_client(armada_svc_endpoint,
self.svc_token)
# Retrieve DeckHand Endpoint Information
deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
# Retrieve last committed revision id
committed_revision_id = self.xcom_puller.get_design_version()
# Get deckhand design reference url
self.deckhand_design_ref = self._init_deckhand_design_ref(
deckhand_svc_endpoint, committed_revision_id)
def get_deckhand_design_ref(self, context):
# Retrieve DeckHand Endpoint Information
svc_type = 'deckhand'
deckhand_svc_endpoint = ucp_service_endpoint(self, svc_type=svc_type)
logging.info("Deckhand endpoint is %s", deckhand_svc_endpoint)
# Retrieve revision_id from xcom
# Note that in the case of 'deploy_site', the dag_id will
# be 'deploy_site.deckhand_get_design_version' for the
# 'deckhand_get_design_version' task. We need to extract
# the xcom value from it in order to get the value of the
# last committed revision ID
committed_revision_id = context['task_instance'].xcom_pull(
task_ids='deckhand_get_design_version',
dag_id=self.main_dag_name + '.deckhand_get_design_version')
# Form Design Reference Path that we will use to retrieve
# the Design YAMLs
deckhand_path = "deckhand+" + deckhand_svc_endpoint
deckhand_design_ref = os.path.join(deckhand_path, "revisions",
str(committed_revision_id),
"rendered-documents")
return deckhand_design_ref
def run_base(self, context):
# Read and parse shiyard.conf
config = configparser.ConfigParser()
config.read(self.shipyard_conf)
# Initialize variables
self.deckhand_client_read_timeout = int(config.get(
'requests_config', 'deckhand_client_read_timeout'))
self.validation_read_timeout = int(config.get(
'requests_config', 'validation_read_timeout'))
# Logs uuid of Shipyard action
LOG.info("Executing Shipyard Action %s",
self.action_info['id'])
# Retrieve Endpoint Information
self.deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
LOG.info("Deckhand endpoint is %s",
self.deckhand_svc_endpoint)
# Set up DeckHand Client
LOG.info("Setting up DeckHand Client...")
# NOTE: The communication between the Airflow workers
# and Deckhand happens via the 'internal' endpoint.
self.deckhandclient = deckhand_client.Client(
session=self.svc_session, endpoint_type='internal')
if not self.deckhandclient:
raise AirflowException('Failed to set up deckhand client!')
def promenade_base(self, context):
# Define task_instance
task_instance = context['task_instance']
# Extract information related to current workflow
# The workflow_info variable will be a dictionary
# that contains information about the workflow such
# as action_id, name and other related parameters
self.workflow_info = task_instance.xcom_pull(task_ids='action_xcom',
key='action',
dag_id=self.main_dag_name)
# Logs uuid of Shipyard action
logging.info("Executing Shipyard Action %s", self.workflow_info['id'])
# Retrieve information of the server that we want to redeploy
# if user executes the 'redeploy_server' dag
if self.workflow_info['dag_id'] == 'redeploy_server':
self.redeploy_server = self.workflow_info['parameters'].get(
'server-name')
if self.redeploy_server:
logging.info("Server to be redeployed is %s",
self.redeploy_server)
else:
raise AirflowException('%s was unable to retrieve the '
'server to be redeployed.' %
self.__class__.__name__)
# Retrieve promenade endpoint
self.promenade_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.promenade_svc_type)
logging.info("Promenade endpoint is %s", self.promenade_svc_endpoint)
def promenade_base(self, context):
# Define task_instance
task_instance = context['task_instance']
# Set up and retrieve values from xcom
self.xcom_puller = XcomPuller(self.main_dag_name, task_instance)
self.action_info = self.xcom_puller.get_action_info()
self.dc = self.xcom_puller.get_deployment_configuration()
# Logs uuid of Shipyard action
logging.info("Executing Shipyard Action %s", self.action_info['id'])
# Retrieve information of the server that we want to redeploy
# if user executes the 'redeploy_server' dag
if self.action_info['dag_id'] == 'redeploy_server':
self.redeploy_server = self.action_info['parameters'].get(
'server-name')
if self.redeploy_server:
logging.info("Server to be redeployed is %s",
self.redeploy_server)
else:
raise AirflowException('%s was unable to retrieve the '
'server to be redeployed.' %
self.__class__.__name__)
# Retrieve promenade endpoint
self.promenade_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.promenade_svc_type)
logging.info("Promenade endpoint is %s", self.promenade_svc_endpoint)
def deckhand_base(self, context):
# Read and parse shiyard.conf
config = configparser.ConfigParser()
config.read(self.shipyard_conf)
# Initialize variables
self.deckhand_client_read_timeout = int(config.get(
'requests_config', 'deckhand_client_read_timeout'))
self.validation_read_timeout = int(config.get(
'requests_config', 'validation_read_timeout'))
# Define task_instance
task_instance = context['task_instance']
# Set up and retrieve values from xcom
self.xcom_puller = XcomPuller(self.main_dag_name, task_instance)
self.action_info = self.xcom_puller.get_action_info()
# Logs uuid of Shipyard action
logging.info("Executing Shipyard Action %s",
self.action_info['id'])
# Retrieve Endpoint Information
self.deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
logging.info("Deckhand endpoint is %s",
self.deckhand_svc_endpoint)
# Set up DeckHand Client
logging.info("Setting up DeckHand Client...")
# NOTE: The communication between the Airflow workers
# and Deckhand happens via the 'internal' endpoint.
self.deckhandclient = deckhand_client.Client(
session=self.svc_session, endpoint_type='internal')
if not self.deckhandclient:
raise AirflowException('Failed to set up deckhand client!')
# Retrieve 'revision_id' from xcom for tasks other than
# 'deckhand_get_design_version'
#
# NOTE: In the case of 'deploy_site', the dag_id will
# be 'deploy_site.deckhand_get_design_version' for the
# 'deckhand_get_design_version' task. We need to extract
# the xcom value from it in order to get the value of the
# last committed revision ID
if self.task_id != 'deckhand_get_design_version':
# Retrieve 'revision_id' from xcom
self.revision_id = self.xcom_puller.get_design_version()
if self.revision_id:
logging.info("Revision ID is %d", self.revision_id)
else:
raise AirflowException('Failed to retrieve Revision ID!')
def run_base(self, context):
# Logs uuid of Shipyard action
LOG.info("Executing Shipyard Action %s", self.action_info['id'])
# Retrieve information of the server that we want to redeploy
# if user executes the 'redeploy_server' dag
if self.action_info['dag_id'] == 'redeploy_server':
self.redeploy_server = self.action_info['parameters'].get(
'server-name')
if self.redeploy_server:
LOG.info("Server to be redeployed is %s", self.redeploy_server)
else:
raise AirflowException('%s was unable to retrieve the '
'server to be redeployed.' %
self.__class__.__name__)
# Retrieve promenade endpoint
self.promenade_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.promenade_svc_type)
LOG.info("Promenade endpoint is %s", self.promenade_svc_endpoint)
# Retrieve Deckhand Endpoint Information
deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
LOG.info("Deckhand endpoint is %s", deckhand_svc_endpoint)
# Form Deckhand Design Reference Path
# This URL will be used to retrieve the Site Design YAMLs
deckhand_path = "deckhand+" + deckhand_svc_endpoint
self.deckhand_design_ref = os.path.join(deckhand_path, "revisions",
str(self.revision_id),
"rendered-documents")
if self.deckhand_design_ref:
LOG.info("Design YAMLs will be retrieved from %s",
self.deckhand_design_ref)
else:
raise AirflowException("Unable to Retrieve Deckhand Revision "
"%d!" % self.revision_id)
def execute(self, context):
# Initialize Variables
context['svc_type'] = 'deckhand'
deckhand_design_version = None
# Define task_instance
task_instance = context['task_instance']
# Extract information related to current workflow
# The workflow_info variable will be a dictionary
# that contains information about the workflow such
# as action_id, name and other related parameters
workflow_info = task_instance.xcom_pull(task_ids='action_xcom',
key='action',
dag_id=self.main_dag_name)
# Logs uuid of action performed by the Operator
logging.info("DeckHand Operator for action %s", workflow_info['id'])
# Retrieve Endpoint Information
context['svc_endpoint'] = ucp_service_endpoint(self, context)
logging.info("Deckhand endpoint is %s", context['svc_endpoint'])
# Deckhand API Call
# Retrieve Design Version from DeckHand
if self.action == 'deckhand_get_design_version':
# Retrieve DeckHand Design Version
deckhand_design_version = self.deckhand_get_design(context)
if deckhand_design_version:
return deckhand_design_version
else:
raise AirflowException('Failed to retrieve revision ID!')
# Validate Design using DeckHand
elif self.action == 'deckhand_validate_site_design':
# Retrieve revision_id from xcom
# Note that in the case of 'deploy_site', the dag_id will be
# 'deploy_site.deckhand_get_design_version.deckhand_get_design_version'
# for the 'deckhand_get_design_version' task. We need to extract
# the xcom value from it in order to get the value of the last
# committed revision ID
context['revision_id'] = task_instance.xcom_pull(
task_ids='deckhand_get_design_version',
dag_id=self.main_dag_name + '.deckhand_get_design_version' * 2)
logging.info("Revision ID is %d", context['revision_id'])
self.deckhand_validate_site(context)
# No action to perform
else:
logging.info('No Action to Perform')
def run_base(self, context):
# Set up xcom_pusher to push values to xcom
self.xcom_pusher = XcomPusher(self.task_instance)
# Logs uuid of action performed by the Operator
LOG.info("Armada Operator for action %s", self.action_info['id'])
# Retrieve Endpoint Information
armada_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.armada_svc_type)
# Set up armada client
self.armada_client = self._init_armada_client(armada_svc_endpoint,
self.svc_token)
# Retrieve DeckHand Endpoint Information
deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
# Get deckhand design reference url
self.deckhand_design_ref = self._init_deckhand_design_ref(
deckhand_svc_endpoint)
def execute(self, context):
# Initialize variable
ucp_components = [
'armada', 'deckhand', 'kubernetesprovisioner',
'physicalprovisioner', 'shipyard'
]
# Define task_instance
self.task_instance = context['task_instance']
# Set up and retrieve values from xcom
self.xcom_puller = XcomPuller(self.main_dag_name, self.task_instance)
self.action_info = self.xcom_puller.get_action_info()
# Set up xcom_pusher to push values to xcom
self.xcom_pusher = XcomPusher(self.task_instance)
# Loop through various UCP Components
for component in ucp_components:
# Retrieve Endpoint Information
service_endpoint = ucp_service_endpoint(self, svc_type=component)
LOG.info("%s endpoint is %s", component, service_endpoint)
# Construct Health Check Endpoint
healthcheck_endpoint = os.path.join(service_endpoint, 'health')
LOG.info("%s healthcheck endpoint is %s", component,
healthcheck_endpoint)
try:
LOG.info("Performing Health Check on %s", component)
# Set health check timeout to 30 seconds
req = requests.get(healthcheck_endpoint, timeout=30)
# An empty response/body returned by a component means
# that it is healthy
if req.status_code == 204:
LOG.info("%s is alive and healthy", component)
except requests.exceptions.RequestException as e:
self.log_health_exception(component, e)
def execute(self, context):
# Initialize variable
ucp_components = [
'armada',
'deckhand',
'kubernetesprovisioner',
'physicalprovisioner',
'shipyard']
# Loop through various UCP Components
for i in ucp_components:
# Retrieve Endpoint Information
service_endpoint = ucp_service_endpoint(self, svc_type=i)
logging.info("%s endpoint is %s", i, service_endpoint)
# Construct Health Check Endpoint
healthcheck_endpoint = os.path.join(service_endpoint,
'health')
logging.info("%s healthcheck endpoint is %s", i,
healthcheck_endpoint)
try:
logging.info("Performing Health Check on %s", i)
# Set health check timeout to 30 seconds
req = requests.get(healthcheck_endpoint, timeout=30)
except requests.exceptions.RequestException as e:
raise AirflowException(e)
# UCP Component will return empty response/body to show that
# it is healthy
if req.status_code == 204:
logging.info("%s is alive and healthy", i)
else:
logging.error(req.text)
raise AirflowException("Invalid Response!")
def run_base(self, context):
# Logs uuid of Shipyard action
LOG.info("Executing Shipyard Action %s", self.action_info['id'])
# Retrieve information of the server that we want to redeploy
# if user executes the 'redeploy_server' dag
if self.action_info['dag_id'] == 'redeploy_server':
self.redeploy_server = self.action_info['parameters'].get(
'server-name')
if self.redeploy_server:
LOG.info("Server to be redeployed is %s", self.redeploy_server)
else:
raise AirflowException('%s was unable to retrieve the '
'server to be redeployed.' %
self.__class__.__name__)
# Retrieve promenade endpoint
self.promenade_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.promenade_svc_type)
LOG.info("Promenade endpoint is %s", self.promenade_svc_endpoint)
def drydock_query_task(self, drydock_client, context, interval,
time_out, task_id):
# Initialize Variables
keystone_token_expired = False
new_dd_client = None
dd_client = drydock_client
# Calculate number of times to execute the 'for' loop
# Convert 'time_out' and 'interval' from string into integer
# The result from the division will be a floating number which
# We will round off to nearest whole number
end_range = round(int(time_out) / int(interval))
# Query task status
for i in range(0, end_range + 1):
if keystone_token_expired:
logging.info("Established new drydock session")
dd_client = new_dd_client
try:
# Retrieve current task state
task_state = dd_client.get_task(task_id=task_id)
task_status = task_state.get('status')
task_result = task_state.get('result')['status']
logging.info("Current status of task id %s is %s",
task_id, task_status)
keystone_token_expired = False
except errors.ClientUnauthorizedError as unauthorized_error:
# TODO: This is a temporary workaround. Drydock will be
# updated with the appropriate fix in the drydock api
# client by having the session detect a 401/403 response
# and refresh the token appropriately.
# Logs drydock client unauthorized error
keystone_token_expired = True
logging.error(unauthorized_error)
# Set up new drydock client with new keystone token
logging.info("Setting up new drydock session...")
drydock_svc_endpoint = ucp_service_endpoint(
self, svc_type='physicalprovisioner')
new_dd_client = self.drydock_session_client(
drydock_svc_endpoint)
except errors.ClientForbiddenError as forbidden_error:
raise AirflowException(forbidden_error)
except errors.ClientError as client_error:
raise AirflowException(client_error)
except:
# There can be instances where there are intermittent network
# issues that prevents us from retrieving the task state. We
# will want to retry in such situations.
logging.info("Unable to retrieve task state. Retrying...")
# Raise Time Out Exception
if task_status == 'running' and i == end_range:
raise AirflowException("Task Execution Timed Out!")
# Exit 'for' loop if the task is in 'complete' or 'terminated'
# state
if task_status in ['complete', 'terminated']:
logging.info('Task result is %s', task_result)
break
else:
time.sleep(int(interval))
# Get final task result
if task_result == 'success':
logging.info('Task id %s has been successfully completed',
self.task_id)
else:
raise AirflowException("Failed to execute/complete task!")
def drydock_base(self, context):
# Initialize Variables
drydock_url = None
dd_session = None
# Define task_instance
task_instance = context['task_instance']
# Set up and retrieve values from xcom
self.xcom_puller = XcomPuller(self.main_dag_name, task_instance)
self.action_info = self.xcom_puller.get_action_info()
self.dc = self.xcom_puller.get_deployment_configuration()
# Logs uuid of action performed by the Operator
logging.info("DryDock Operator for action %s", self.action_info['id'])
# Retrieve information of the server that we want to redeploy if user
# executes the 'redeploy_server' dag
# Set node filter to be the server that we want to redeploy
if self.action_info['dag_id'] == 'redeploy_server':
self.redeploy_server = (
self.action_info['parameters']['server-name'])
if self.redeploy_server:
logging.info("Server to be redeployed is %s",
self.redeploy_server)
self.node_filter = self.redeploy_server
else:
raise AirflowException('Unable to retrieve information of '
'node to be redeployed!')
# Retrieve Endpoint Information
self.drydock_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.drydock_svc_type)
logging.info("Drydock endpoint is %s", self.drydock_svc_endpoint)
# Parse DryDock Service Endpoint
drydock_url = urlparse(self.drydock_svc_endpoint)
# Build a DrydockSession with credentials and target host
# information.
# The DrydockSession will care for TCP connection pooling
# and header management
logging.info("Build DryDock Session")
dd_session = session.DrydockSession(drydock_url.hostname,
port=drydock_url.port,
auth_gen=self._auth_gen)
# Raise Exception if we are not able to set up the session
if dd_session:
logging.info("Successfully Set Up DryDock Session")
else:
raise AirflowException("Failed to set up Drydock Session!")
# Use the DrydockSession to build a DrydockClient that can
# be used to make one or more API calls
logging.info("Create DryDock Client")
self.drydock_client = client.DrydockClient(dd_session)
# Raise Exception if we are not able to build the client
if self.drydock_client:
logging.info("Successfully Set Up DryDock client")
else:
raise AirflowException("Failed to set up Drydock Client!")
# Retrieve DeckHand Endpoint Information
deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
logging.info("Deckhand endpoint is %s", deckhand_svc_endpoint)
# Retrieve last committed revision id
committed_revision_id = self.xcom_puller.get_design_version()
# Form DeckHand Design Reference Path
# This URL will be used to retrieve the Site Design YAMLs
deckhand_path = "deckhand+" + deckhand_svc_endpoint
self.deckhand_design_ref = os.path.join(deckhand_path, "revisions",
str(committed_revision_id),
"rendered-documents")
if self.deckhand_design_ref:
logging.info("Design YAMLs will be retrieved from %s",
self.deckhand_design_ref)
else:
raise AirflowException("Unable to Retrieve Design Reference!")
def execute(self, context):
# Initialize Variables
armada_client = None
design_ref = None
# Define task_instance
task_instance = context['task_instance']
# Extract information related to current workflow
# The workflow_info variable will be a dictionary
# that contains information about the workflow such
# as action_id, name and other related parameters
workflow_info = task_instance.xcom_pull(task_ids='action_xcom',
key='action',
dag_id=self.main_dag_name)
# Logs uuid of action performed by the Operator
logging.info("Armada Operator for action %s", workflow_info['id'])
# Retrieve Deckhand Design Reference
design_ref = self.get_deckhand_design_ref(context)
if design_ref:
logging.info("Design YAMLs will be retrieved from %s", design_ref)
else:
raise AirflowException("Unable to Retrieve Design Reference!")
# Validate Site Design
if self.action == 'validate_site_design':
# Initialize variable
armada_svc_endpoint = None
site_design_validity = 'invalid'
# Retrieve Endpoint Information
svc_type = 'armada'
armada_svc_endpoint = ucp_service_endpoint(self, svc_type=svc_type)
site_design_validity = self.armada_validate_site_design(
armada_svc_endpoint, design_ref)
if site_design_validity == 'valid':
logging.info("Site Design has been successfully validated")
else:
raise AirflowException("Site Design Validation Failed!")
return site_design_validity
# Create Armada Client
# Retrieve Endpoint Information
svc_type = 'armada'
armada_svc_endpoint = ucp_service_endpoint(self, svc_type=svc_type)
logging.info("Armada endpoint is %s", armada_svc_endpoint)
# Set up Armada Client
armada_client = self.armada_session_client(armada_svc_endpoint)
# Retrieve Tiller Information and assign to context 'query'
context['query'] = self.get_tiller_info(context)
# Armada API Call
# Armada Status
if self.action == 'armada_status':
self.get_armada_status(context, armada_client)
# Armada Apply
elif self.action == 'armada_apply':
# TODO (bryan-strassner) externalize the name of the manifest to
# use this needs to come from a site configuration document for
# consumption by shipyard/airflow. For now. "full-site" is the
# only value that will work.
target_manifest = 'full-site'
self.armada_apply(context, armada_client, design_ref,
target_manifest)
# Armada Get Releases
elif self.action == 'armada_get_releases':
self.armada_get_releases(context, armada_client)
else:
logging.info('No Action to Perform')
def execute(self, context):
# Initialize Variable
redeploy_server = None
# Placeholder definition
# TODO: Need to decide how to pass the required value from Shipyard to
# the 'node_filter' variable. No filter will be used for now.
self.node_filter = None
# Define task_instance
task_instance = context['task_instance']
# Extract information related to current workflow
# The workflow_info variable will be a dictionary
# that contains information about the workflow such
# as action_id, name and other related parameters
workflow_info = task_instance.xcom_pull(
task_ids='action_xcom', key='action',
dag_id=self.main_dag_name)
# Logs uuid of action performed by the Operator
logging.info("DryDock Operator for action %s", workflow_info['id'])
# Retrieve information of the server that we want to redeploy if user
# executes the 'redeploy_server' dag
# Set node filter to be the server that we want to redeploy
if workflow_info['dag_id'] == 'redeploy_server':
redeploy_server = workflow_info['parameters'].get('server-name')
if redeploy_server:
logging.info("Server to be redeployed is %s", redeploy_server)
self.node_filter = redeploy_server
else:
raise AirflowException('Unable to retrieve information of '
'node to be redeployed!')
# Retrieve Deckhand Design Reference
self.design_ref = self.get_deckhand_design_ref(context)
if self.design_ref:
logging.info("Drydock YAMLs will be retrieved from %s",
self.design_ref)
else:
raise AirflowException("Unable to Retrieve Design Reference!")
# Drydock Validate Site Design
if self.action == 'validate_site_design':
# Initialize variable
site_design_validity = 'invalid'
# Retrieve Endpoint Information
svc_type = 'physicalprovisioner'
drydock_svc_endpoint = ucp_service_endpoint(self,
svc_type=svc_type)
site_design_validity = self.drydock_validate_design(
drydock_svc_endpoint)
return site_design_validity
# DrydockClient
# Retrieve Endpoint Information
svc_type = 'physicalprovisioner'
drydock_svc_endpoint = ucp_service_endpoint(self,
svc_type=svc_type)
logging.info("DryDock endpoint is %s", drydock_svc_endpoint)
# Set up DryDock Client
drydock_client = self.drydock_session_client(drydock_svc_endpoint)
# Read shipyard.conf
config = configparser.ConfigParser()
config.read(self.shipyard_conf)
if not config.read(self.shipyard_conf):
raise AirflowException("Unable to read content of shipyard.conf")
# Create Task for verify_site
if self.action == 'verify_site':
# Default settings for 'verify_site' execution is to query
# the task every 10 seconds and to time out after 60 seconds
query_interval = config.get('drydock',
'verify_site_query_interval')
task_timeout = config.get('drydock', 'verify_site_task_timeout')
self.drydock_action(drydock_client, context, self.action,
query_interval, task_timeout)
# Create Task for prepare_site
elif self.action == 'prepare_site':
# Default settings for 'prepare_site' execution is to query
# the task every 10 seconds and to time out after 300 seconds
query_interval = config.get('drydock',
'prepare_site_query_interval')
task_timeout = config.get('drydock', 'prepare_site_task_timeout')
self.drydock_action(drydock_client, context, self.action,
query_interval, task_timeout)
# Create Task for prepare_node
elif self.action == 'prepare_nodes':
# Default settings for 'prepare_node' execution is to query
# the task every 30 seconds and to time out after 1800 seconds
query_interval = config.get('drydock',
'prepare_node_query_interval')
task_timeout = config.get('drydock', 'prepare_node_task_timeout')
self.drydock_action(drydock_client, context, self.action,
query_interval, task_timeout)
# Create Task for deploy_node
elif self.action == 'deploy_nodes':
# Default settings for 'deploy_node' execution is to query
# the task every 30 seconds and to time out after 3600 seconds
query_interval = config.get('drydock',
'deploy_node_query_interval')
task_timeout = config.get('drydock', 'deploy_node_task_timeout')
self.drydock_action(drydock_client, context, self.action,
query_interval, task_timeout)
# Wait for 120 seconds (default value) before checking the cluster
# join process as it takes time for process to be triggered across
# all nodes
cluster_join_check_backoff_time = config.get(
'drydock', 'cluster_join_check_backoff_time')
logging.info("All nodes deployed in MAAS")
logging.info("Wait for %d seconds before checking node state...",
int(cluster_join_check_backoff_time))
time.sleep(int(cluster_join_check_backoff_time))
# Check that cluster join process is completed before declaring
# deploy_node as 'completed'. Set time out to 30 minutes and set
# polling interval to 30 seconds.
check_node_status(1800, 30)
# Create Task for destroy_node
# NOTE: This is a PlaceHolder function. The 'destroy_node'
# functionalities in DryDock is being worked on and is not
# ready at the moment.
elif self.action == 'destroy_node':
# Default settings for 'destroy_node' execution is to query
# the task every 30 seconds and to time out after 900 seconds
query_interval = config.get('drydock',
'destroy_node_query_interval')
task_timeout = config.get('drydock', 'destroy_node_task_timeout')
logging.info("Destroying node %s from cluster...", redeploy_server)
time.sleep(15)
logging.info("Successfully deleted node %s", redeploy_server)
# TODO: Uncomment when the function to destroy/delete node is
# ready for consumption in Drydock
# self.drydock_action(drydock_client, context, self.action,
# query_interval, task_timeout)
# Do not perform any action
else:
logging.info('No Action to Perform')
def run_base(self, context):
# Logs uuid of action performed by the Operator
LOG.info("DryDock Operator for action %s", self.action_info['id'])
# Skip workflow if health checks on Drydock failed and continue-on-fail
# option is turned on
if self.xcom_puller.get_check_drydock_continue_on_fail():
LOG.info(
"Skipping %s as health checks on Drydock have "
"failed and continue-on-fail option has been "
"turned on", self.__class__.__name__)
# Set continue processing to False
self.continue_processing = False
return
# Retrieve information of the server that we want to redeploy if user
# executes the 'redeploy_server' dag
# Set node filter to be the server that we want to redeploy
if self.action_info['dag_id'] == 'redeploy_server':
self.redeploy_server = (
self.action_info['parameters']['server-name'])
if self.redeploy_server:
LOG.info("Server to be redeployed is %s", self.redeploy_server)
self.node_filter = self.redeploy_server
else:
raise AirflowException('%s was unable to retrieve the '
'server to be redeployed.' %
self.__class__.__name__)
# Retrieve Endpoint Information
self.drydock_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.drydock_svc_type)
LOG.info("Drydock endpoint is %s", self.drydock_svc_endpoint)
# Parse DryDock Service Endpoint
drydock_url = urlparse(self.drydock_svc_endpoint)
# Build a DrydockSession with credentials and target host
# information.
# The DrydockSession will care for TCP connection pooling
# and header management
LOG.info("Build DryDock Session")
dd_session = session.DrydockSession(drydock_url.hostname,
port=drydock_url.port,
auth_gen=self._auth_gen)
# Raise Exception if we are not able to set up the session
if dd_session:
LOG.info("Successfully Set Up DryDock Session")
else:
raise AirflowException("Failed to set up Drydock Session!")
# Use the DrydockSession to build a DrydockClient that can
# be used to make one or more API calls
LOG.info("Create DryDock Client")
self.drydock_client = client.DrydockClient(dd_session)
# Raise Exception if we are not able to build the client
if self.drydock_client:
LOG.info("Successfully Set Up DryDock client")
else:
raise AirflowException("Failed to set up Drydock Client!")
# Retrieve DeckHand Endpoint Information
deckhand_svc_endpoint = ucp_service_endpoint(
self, svc_type=self.deckhand_svc_type)
LOG.info("Deckhand endpoint is %s", deckhand_svc_endpoint)
# Form DeckHand Design Reference Path
# This URL will be used to retrieve the Site Design YAMLs
deckhand_path = "deckhand+" + deckhand_svc_endpoint
self.deckhand_design_ref = os.path.join(deckhand_path, "revisions",
str(self.revision_id),
"rendered-documents")
if self.deckhand_design_ref:
LOG.info("Design YAMLs will be retrieved from %s",
self.deckhand_design_ref)
else:
raise AirflowException("Unable to Retrieve Design Reference!")