"""Carry out alarming requests via Aodh API."""
import json
-
import logging
-from core.message_bus.producer import KafkaProducer
-
-from plugins.OpenStack.response import OpenStack_Response
-from plugins.OpenStack.settings import Config
+import six
+import yaml
-__author__ = "Helena McGough"
+from osm_mon.core.auth import AuthManager
+from osm_mon.core.database import DatabaseManager
+from osm_mon.core.message_bus.producer import KafkaProducer
+from osm_mon.core.settings import Config
+from osm_mon.plugins.OpenStack.Gnocchi.metrics import METRIC_MAPPINGS
+from osm_mon.plugins.OpenStack.common import Common
+from osm_mon.plugins.OpenStack.response import OpenStack_Response
log = logging.getLogger(__name__)
-ALARM_NAMES = {
- "average_memory_usage_above_threshold": "average_memory_utilization",
- "disk_read_ops": "disk_read_ops",
- "disk_write_ops": "disk_write_ops",
- "disk_read_bytes": "disk_read_bytes",
- "disk_write_bytes": "disk_write_bytes",
- "net_packets_dropped": "packets_dropped",
- "packets_in_above_threshold": "packets_received",
- "packets_out_above_threshold": "packets_sent",
- "cpu_utilization_above_threshold": "cpu_utilization"}
-
SEVERITIES = {
"warning": "low",
"minor": "low",
"indeterminate": "critical"}
STATISTICS = {
- "average": "avg",
+ "average": "mean",
"minimum": "min",
"maximum": "max",
"count": "count",
def __init__(self):
"""Create the OpenStack alarming instance."""
- # Initialize configuration and notifications
- config = Config.instance()
- config.read_environ("aodh")
-
- # Initialise authentication for API requests
- self.auth_token = None
- self.endpoint = None
- self.common = None
+ self._database_manager = DatabaseManager()
+ self._auth_manager = AuthManager()
# Use the Response class to generate valid json response messages
self._response = OpenStack_Response()
# Initializer a producer to send responses back to SO
self._producer = KafkaProducer("alarm_response")
- def alarming(self, message, common, auth_token):
- """Consume info from the message bus to manage alarms."""
- values = json.loads(message.value)
- self.common = common
+ def alarming(self, message, vim_uuid):
+ """
+ Processes alarm request message depending on it's key
+ :param message: Message containing key and value attributes. This last one can be in JSON or YAML format.
+ :param vim_uuid: UUID of the VIM to handle the alarm request.
+ :return:
+ """
+ try:
+ values = json.loads(message.value)
+ except ValueError:
+ values = yaml.safe_load(message.value)
log.info("OpenStack alarm action required.")
- # Generate and auth_token and endpoint for request
- if auth_token is not None:
- if self.auth_token != auth_token:
- log.info("Auth_token for alarming set by access_credentials.")
- self.auth_token = auth_token
- else:
- log.info("Auth_token has not been updated.")
- else:
- log.info("Using environment variables to set auth_token for Aodh.")
- self.auth_token = self.common._authenticate()
+ verify_ssl = self._auth_manager.is_verify_ssl(vim_uuid)
- if self.endpoint is None:
- log.info("Generating a new endpoint for Aodh.")
- self.endpoint = self.common.get_endpoint("alarming")
+ auth_token = Common.get_auth_token(vim_uuid, verify_ssl=verify_ssl)
- if message.key == "create_alarm_request":
- # Configure/Update an alarm
- alarm_details = values['alarm_create_request']
+ alarm_endpoint = Common.get_endpoint("alarming", vim_uuid, verify_ssl=verify_ssl)
+ metric_endpoint = Common.get_endpoint("metric", vim_uuid, verify_ssl=verify_ssl)
- alarm_id, alarm_status = self.configure_alarm(
- self.endpoint, self.auth_token, alarm_details)
+ vim_account = self._auth_manager.get_credentials(vim_uuid)
+ vim_config = json.loads(vim_account.config)
- # Generate a valid response message, send via producer
+ if message.key == "create_alarm_request":
+ alarm_details = values['alarm_create_request']
+ alarm_id = None
+ status = False
try:
- if alarm_status is True:
- log.info("Alarm successfully created")
-
- resp_message = self._response.generate_response(
- 'create_alarm_response', status=alarm_status,
- alarm_id=alarm_id,
- cor_id=alarm_details['correlation_id'])
- log.info("Response Message: %s", resp_message)
- self._producer.create_alarm_response(
- 'create_alarm_response', resp_message,
- 'alarm_response')
- except Exception as exc:
- log.warn("Response creation failed: %s", exc)
+ metric_name = alarm_details['metric_name'].lower()
+ resource_id = alarm_details['resource_uuid']
+
+ self.check_for_metric(auth_token, metric_endpoint, metric_name, resource_id, verify_ssl)
+
+ alarm_id = self.configure_alarm(
+ alarm_endpoint, auth_token, alarm_details, vim_config, verify_ssl)
+
+ log.info("Alarm successfully created")
+ self._database_manager.save_alarm(alarm_id,
+ vim_uuid,
+ alarm_details['threshold_value'],
+ alarm_details['operation'].lower(),
+ alarm_details['metric_name'].lower(),
+ alarm_details['vdu_name'].lower(),
+ alarm_details['vnf_member_index'],
+ alarm_details['ns_id'].lower()
+ )
+ status = True
+ except Exception as e:
+ log.exception("Error creating alarm")
+ raise e
+ finally:
+ self._generate_and_send_response('create_alarm_response',
+ alarm_details['correlation_id'],
+ status=status,
+ alarm_id=alarm_id)
elif message.key == "list_alarm_request":
- # Check for a specifed: alarm_name, resource_uuid, severity
- # and generate the appropriate list
list_details = values['alarm_list_request']
-
- alarm_list = self.list_alarms(
- self.endpoint, self.auth_token, list_details)
-
+ alarm_list = None
try:
- # Generate and send a list response back
- resp_message = self._response.generate_response(
- 'list_alarm_response', alarm_list=alarm_list,
- cor_id=list_details['correlation_id'])
- log.info("Response Message: %s", resp_message)
- self._producer.list_alarm_response(
- 'list_alarm_response', resp_message,
- 'alarm_response')
- except Exception as exc:
- log.warn("Failed to send a valid response back.")
+ alarm_list = self.list_alarms(
+ alarm_endpoint, auth_token, list_details, verify_ssl)
+ except Exception as e:
+ log.exception("Error listing alarms")
+ raise e
+ finally:
+ self._generate_and_send_response('list_alarm_response',
+ list_details['correlation_id'],
+ alarm_list=alarm_list)
elif message.key == "delete_alarm_request":
request_details = values['alarm_delete_request']
alarm_id = request_details['alarm_uuid']
-
- resp_status = self.delete_alarm(
- self.endpoint, self.auth_token, alarm_id)
-
- # Generate and send a response message
+ status = False
try:
- resp_message = self._response.generate_response(
- 'delete_alarm_response', alarm_id=alarm_id,
- status=resp_status,
- cor_id=request_details['correlation_id'])
- log.info("Response message: %s", resp_message)
- self._producer.delete_alarm_response(
- 'delete_alarm_response', resp_message,
- 'alarm_response')
- except Exception as exc:
- log.warn("Failed to create delete reponse:%s", exc)
+ self.delete_alarm(
+ alarm_endpoint, auth_token, alarm_id, verify_ssl)
+ status = True
+ except Exception as e:
+ log.exception("Error deleting alarm")
+ raise e
+ finally:
+ self._generate_and_send_response('delete_alarm_response',
+ request_details['correlation_id'],
+ status=status,
+ alarm_id=alarm_id)
elif message.key == "acknowledge_alarm":
- # Acknowledge that an alarm has been dealt with by the SO
- alarm_id = values['ack_details']['alarm_uuid']
+ try:
+ alarm_id = values['ack_details']['alarm_uuid']
- response = self.update_alarm_state(
- self.endpoint, self.auth_token, alarm_id)
+ self.update_alarm_state(
+ alarm_endpoint, auth_token, alarm_id, verify_ssl)
- # Log if an alarm was reset
- if response is True:
log.info("Acknowledged the alarm and cleared it.")
- else:
- log.warn("Failed to acknowledge/clear the alarm.")
+ except Exception as e:
+ log.exception("Error acknowledging alarm")
+ raise e
elif message.key == "update_alarm_request":
# Update alarm configurations
alarm_details = values['alarm_update_request']
-
- alarm_id, status = self.update_alarm(
- self.endpoint, self.auth_token, alarm_details)
-
- # Generate a response for an update request
+ alarm_id = None
+ status = False
try:
- resp_message = self._response.generate_response(
- 'update_alarm_response', alarm_id=alarm_id,
- cor_id=alarm_details['correlation_id'],
- status=status)
- log.info("Response message: %s", resp_message)
- self._producer.update_alarm_response(
- 'update_alarm_response', resp_message,
- 'alarm_response')
- except Exception as exc:
- log.warn("Failed to send an update response:%s", exc)
+ alarm_id = self.update_alarm(
+ alarm_endpoint, auth_token, alarm_details, vim_config, verify_ssl)
+ status = True
+ except Exception as e:
+ log.exception("Error updating alarm")
+ raise e
+ finally:
+ self._generate_and_send_response('update_alarm_response',
+ alarm_details['correlation_id'],
+ status=status,
+ alarm_id=alarm_id)
else:
log.debug("Unknown key, no action will be performed")
- return
-
- def configure_alarm(self, endpoint, auth_token, values):
+ def configure_alarm(self, alarm_endpoint, auth_token, values, vim_config, verify_ssl):
"""Create requested alarm in Aodh."""
- url = "{}/v2/alarms/".format(endpoint)
+ url = "{}/v2/alarms/".format(alarm_endpoint)
# Check if the desired alarm is supported
alarm_name = values['alarm_name'].lower()
metric_name = values['metric_name'].lower()
resource_id = values['resource_uuid']
- if alarm_name not in ALARM_NAMES.keys():
- log.warn("This alarm is not supported, by a valid metric.")
- return None, False
- if ALARM_NAMES[alarm_name] != metric_name:
- log.warn("This is not the correct metric for this alarm.")
- return None, False
+ if metric_name not in METRIC_MAPPINGS.keys():
+ raise KeyError("Metric {} is not supported.".format(metric_name))
- # Check for the required metric
- metric_id = self.check_for_metric(auth_token, metric_name, resource_id)
-
- try:
- if metric_id is not None:
- # Create the alarm if metric is available
- payload = self.check_payload(values, metric_name, resource_id,
- alarm_name)
- new_alarm = self.common._perform_request(
- url, auth_token, req_type="post", payload=payload)
- return json.loads(new_alarm.text)['alarm_id'], True
- else:
- log.warn("The required Gnocchi metric does not exist.")
- return None, False
-
- except Exception as exc:
- log.warn("Failed to create the alarm: %s", exc)
- return None, False
+ if 'granularity' in vim_config and 'granularity' not in values:
+ values['granularity'] = vim_config['granularity']
+ payload = self.check_payload(values, metric_name, resource_id,
+ alarm_name)
+ new_alarm = Common.perform_request(
+ url, auth_token, req_type="post", payload=payload, verify_ssl=verify_ssl)
+ return json.loads(new_alarm.text)['alarm_id']
- def delete_alarm(self, endpoint, auth_token, alarm_id):
+ def delete_alarm(self, endpoint, auth_token, alarm_id, verify_ssl):
"""Delete alarm function."""
- url = "{}/v2/alarms/%s".format(endpoint) % (alarm_id)
+ url = "{}/v2/alarms/%s".format(endpoint) % alarm_id
- try:
- result = self.common._perform_request(
- url, auth_token, req_type="delete")
- if str(result.status_code) == "404":
- log.info("Alarm doesn't exist: %s", result.status_code)
- # If status code is 404 alarm did not exist
- return False
- else:
- return True
-
- except Exception as exc:
- log.warn("Failed to delete alarm: %s because %s.", alarm_id, exc)
- return False
+ result = Common.perform_request(
+ url, auth_token, req_type="delete", verify_ssl=verify_ssl)
+ if str(result.status_code) == "404":
+ raise ValueError("Alarm {} doesn't exist".format(alarm_id))
- def list_alarms(self, endpoint, auth_token, list_details):
+ def list_alarms(self, endpoint, auth_token, list_details, verify_ssl):
"""Generate the requested list of alarms."""
url = "{}/v2/alarms/".format(endpoint)
a_list, name_list, sev_list, res_list = [], [], [], []
# TODO(mcgoughh): for now resource_id is a mandatory field
- # Check for a reqource is
+ # Check for a resource id
try:
resource = list_details['resource_uuid']
- except KeyError as exc:
- log.warn("Resource id not specified for list request: %s", exc)
- return None
-
- # Checking what fields are specified for a list request
- try:
name = list_details['alarm_name'].lower()
- if name not in ALARM_NAMES.keys():
- log.warn("This alarm is not supported, won't be used!")
- name = None
- except KeyError as exc:
- log.info("Alarm name isn't specified.")
- name = None
-
- try:
severity = list_details['severity'].lower()
sev = SEVERITIES[severity]
- except KeyError as exc:
- log.info("Severity is unspecified/incorrectly configured")
- sev = None
+ except KeyError as e:
+ log.warning("Missing parameter for alarm list request: %s", e)
+ raise e
# Perform the request to get the desired list
try:
- result = self.common._perform_request(
- url, auth_token, req_type="get")
+ result = Common.perform_request(
+ url, auth_token, req_type="get", verify_ssl=verify_ssl)
if result is not None:
# Get list based on resource id
for alarm in json.loads(result.text):
rule = alarm['gnocchi_resources_threshold_rule']
if resource == rule['resource_id']:
- res_list.append(str(alarm))
- if not res_list:
- log.info("No alarms for this resource")
- return a_list
+ res_list.append(alarm['alarm_id'])
# Generate specified listed if requested
if name is not None and sev is not None:
name, sev)
for alarm in json.loads(result.text):
if name == alarm['name']:
- name_list.append(str(alarm))
+ name_list.append(alarm['alarm_id'])
for alarm in json.loads(result.text):
if sev == alarm['severity']:
- sev_list.append(str(alarm))
+ sev_list.append(alarm['alarm_id'])
name_sev_list = list(set(name_list).intersection(sev_list))
a_list = list(set(name_sev_list).intersection(res_list))
elif name is not None:
log.info("Returning a %s list of alarms.", name)
for alarm in json.loads(result.text):
if name == alarm['name']:
- name_list.append(str(alarm))
+ name_list.append(alarm['alarm_id'])
a_list = list(set(name_list).intersection(res_list))
elif sev is not None:
log.info("Returning %s severity alarm list.", sev)
for alarm in json.loads(result.text):
if sev == alarm['severity']:
- sev_list.append(str(alarm))
+ sev_list.append(alarm['alarm_id'])
a_list = list(set(sev_list).intersection(res_list))
else:
log.info("Returning an entire list of alarms.")
a_list = res_list
else:
log.info("There are no alarms!")
+ response_list = []
+ for alarm in json.loads(result.text):
+ if alarm['alarm_id'] in a_list:
+ response_list.append(alarm)
+ return response_list
- except Exception as exc:
- log.info("Failed to generate required list: %s", exc)
- return None
-
- return a_list
+ except Exception as e:
+ log.exception("Failed to generate alarm list: ")
+ raise e
- def update_alarm_state(self, endpoint, auth_token, alarm_id):
+ def update_alarm_state(self, endpoint, auth_token, alarm_id, verify_ssl):
"""Set the state of an alarm to ok when ack message is received."""
url = "{}/v2/alarms/%s/state".format(endpoint) % alarm_id
payload = json.dumps("ok")
- try:
- self.common._perform_request(
- url, auth_token, req_type="put", payload=payload)
- return True
- except Exception as exc:
- log.warn("Unable to update alarm state: %s", exc)
- return False
-
- def update_alarm(self, endpoint, auth_token, values):
+ Common.perform_request(
+ url, auth_token, req_type="put", payload=payload, verify_ssl=verify_ssl)
+
+ def update_alarm(self, endpoint, auth_token, values, vim_config, verify_ssl):
"""Get alarm name for an alarm configuration update."""
# Get already existing alarm details
url = "{}/v2/alarms/%s".format(endpoint) % values['alarm_uuid']
# Gets current configurations about the alarm
- try:
- result = self.common._perform_request(
- url, auth_token, req_type="get")
- alarm_name = json.loads(result.text)['name']
- rule = json.loads(result.text)['gnocchi_resources_threshold_rule']
- alarm_state = json.loads(result.text)['state']
- resource_id = rule['resource_id']
- metric_name = rule['metric']
- except Exception as exc:
- log.warn("Failed to retreive existing alarm info: %s.\
- Can only update OSM alarms.", exc)
- return None, False
+ result = Common.perform_request(
+ url, auth_token, req_type="get")
+ alarm_name = json.loads(result.text)['name']
+ rule = json.loads(result.text)['gnocchi_resources_threshold_rule']
+ alarm_state = json.loads(result.text)['state']
+ resource_id = rule['resource_id']
+ metric_name = [key for key, value in six.iteritems(METRIC_MAPPINGS) if value == rule['metric']][0]
# Generates and check payload configuration for alarm update
+ if 'granularity' in vim_config and 'granularity' not in values:
+ values['granularity'] = vim_config['granularity']
payload = self.check_payload(values, metric_name, resource_id,
alarm_name, alarm_state=alarm_state)
# Updates the alarm configurations with the valid payload
- if payload is not None:
- try:
- update_alarm = self.common._perform_request(
- url, auth_token, req_type="put", payload=payload)
+ update_alarm = Common.perform_request(
+ url, auth_token, req_type="put", payload=payload, verify_ssl=verify_ssl)
- return json.loads(update_alarm.text)['alarm_id'], True
- except Exception as exc:
- log.warn("Alarm update could not be performed: %s", exc)
- return None, False
- return None, False
+ return json.loads(update_alarm.text)['alarm_id']
def check_payload(self, values, metric_name, resource_id,
alarm_name, alarm_state=None):
"""Check that the payload is configuration for update/create alarm."""
- try:
- # Check state and severity
+ cfg = Config.instance()
+ # Check state and severity
+
+ severity = 'critical'
+ if 'severity' in values:
severity = values['severity'].lower()
- if severity == "indeterminate":
- alarm_state = "insufficient data"
- if alarm_state is None:
- alarm_state = "ok"
-
- statistic = values['statistic'].lower()
- # Try to configure the payload for the update/create request
- # Can only update: threshold, operation, statistic and
- # the severity of the alarm
- rule = {'threshold': values['threshold_value'],
- 'comparison_operator': values['operation'].lower(),
- 'metric': metric_name,
- 'resource_id': resource_id,
- 'resource_type': 'generic',
- 'aggregation_method': STATISTICS[statistic], }
- payload = json.dumps({'state': alarm_state,
- 'name': alarm_name,
- 'severity': SEVERITIES[severity],
- 'type': 'gnocchi_resources_threshold',
- 'gnocchi_resources_threshold_rule': rule,
- 'alarm_actions': ['http://localhost:8662'], })
- return payload
- except KeyError as exc:
- log.warn("Alarm is not configured correctly: %s", exc)
- return None
+
+ if severity == "indeterminate":
+ alarm_state = "insufficient data"
+ if alarm_state is None:
+ alarm_state = "ok"
+
+ statistic = values['statistic'].lower()
+
+ granularity = cfg.OS_DEFAULT_GRANULARITY
+ if 'granularity' in values:
+ granularity = values['granularity']
+
+ resource_type = 'generic'
+ if 'resource_type' in values:
+ resource_type = values['resource_type'].lower()
+
+ # Try to configure the payload for the update/create request
+ # Can only update: threshold, operation, statistic and
+ # the severity of the alarm
+ rule = {'threshold': values['threshold_value'],
+ 'comparison_operator': values['operation'].lower(),
+ 'metric': METRIC_MAPPINGS[metric_name],
+ 'resource_id': resource_id,
+ 'resource_type': resource_type,
+ 'aggregation_method': STATISTICS[statistic],
+ 'granularity': granularity, }
+ payload = json.dumps({'state': alarm_state,
+ 'name': alarm_name,
+ 'severity': SEVERITIES[severity],
+ 'type': 'gnocchi_resources_threshold',
+ 'gnocchi_resources_threshold_rule': rule,
+ 'alarm_actions': [cfg.OS_NOTIFIER_URI], }, sort_keys=True)
+ return payload
def get_alarm_state(self, endpoint, auth_token, alarm_id):
"""Get the state of the alarm."""
url = "{}/v2/alarms/%s/state".format(endpoint) % alarm_id
+ alarm_state = Common.perform_request(
+ url, auth_token, req_type="get")
+ return json.loads(alarm_state.text)
+
+ def check_for_metric(self, auth_token, metric_endpoint, metric_name, resource_id, verify_ssl):
+ """
+ Checks if resource has a specific metric. If not, throws exception.
+ :param auth_token: OpenStack auth token
+ :param metric_endpoint: OpenStack metric endpoint
+ :param metric_name: Metric name
+ :param resource_id: Resource UUID
+ :return: Metric details from resource
+ :raise Exception: Could not retrieve metric from resource
+ """
try:
- alarm_state = self.common._perform_request(
- url, auth_token, req_type="get")
- return json.loads(alarm_state.text)
- except Exception as exc:
- log.warn("Failed to get the state of the alarm:%s", exc)
- return None
-
- def check_for_metric(self, auth_token, m_name, r_id):
- """Check for the alarm metric."""
+ url = "{}/v1/resource/generic/{}".format(metric_endpoint, resource_id)
+ result = Common.perform_request(
+ url, auth_token, req_type="get", verify_ssl=verify_ssl)
+ resource = json.loads(result.text)
+ metrics_dict = resource['metrics']
+ return metrics_dict[METRIC_MAPPINGS[metric_name]]
+ except Exception as e:
+ log.exception("Desired Gnocchi metric not found:", e)
+ raise e
+
+ def _generate_and_send_response(self, key, correlation_id, **kwargs):
try:
- endpoint = self.common.get_endpoint("metric")
-
- url = "{}/v1/metric/".format(endpoint)
- metric_list = self.common._perform_request(
- url, auth_token, req_type="get")
-
- for metric in json.loads(metric_list.text):
- name = metric['name']
- resource = metric['resource_id']
- if (name == m_name and resource == r_id):
- metric_id = metric['id']
- log.info("The required metric exists, an alarm will be created.")
- return metric_id
- except Exception as exc:
- log.info("Desired Gnocchi metric not found:%s", exc)
- return None
+ resp_message = self._response.generate_response(
+ key, cor_id=correlation_id, **kwargs)
+ log.info("Response Message: %s", resp_message)
+ self._producer.publish_alarm_response(
+ key, resp_message)
+ except Exception as e:
+ log.exception("Response creation failed:")
+ raise e