X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;f=osm_mon%2Fplugins%2FOpenStack%2FAodh%2Falarming.py;h=13b654162fb7c4a54a791b799eaef0038999c2c7;hb=ffe5a8e4e35d3fc7860eb9231618dc0e84e448b7;hp=234337226b51b86a84af3a9bf0938586212eb9db;hpb=c7397b95dbaeebd7d872779eec809daed9e487cc;p=osm%2FMON.git diff --git a/osm_mon/plugins/OpenStack/Aodh/alarming.py b/osm_mon/plugins/OpenStack/Aodh/alarming.py index 2343372..13b6541 100644 --- a/osm_mon/plugins/OpenStack/Aodh/alarming.py +++ b/osm_mon/plugins/OpenStack/Aodh/alarming.py @@ -22,29 +22,21 @@ """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", @@ -53,7 +45,7 @@ SEVERITIES = { "indeterminate": "critical"} STATISTICS = { - "average": "avg", + "average": "mean", "minimum": "min", "maximum": "max", "count": "count", @@ -67,12 +59,10 @@ class Alarming(object): """Create the OpenStack alarming instance.""" # Initialize configuration and notifications config = Config.instance() - config.read_environ("aodh") + config.read_environ() - # 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() @@ -80,58 +70,94 @@ class Alarming(object): # Initializer a producer to send responses back to SO self._producer = KafkaProducer("alarm_response") - def alarming(self, message, common, auth_token): + def configure_alarm(self, alarm_endpoint, metric_endpoint, auth_token, values, vim_config): + """Create requested alarm in Aodh.""" + 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 metric_name not in METRIC_MAPPINGS.keys(): + log.warning("This metric is not supported.") + return None, False + + # Check for the required metric + metric_id = self.check_for_metric(auth_token, metric_endpoint, metric_name, resource_id) + + try: + if metric_id is not None: + # Create the alarm if metric is available + 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) + return json.loads(new_alarm.text)['alarm_id'], True + else: + log.warning("The required Gnocchi metric does not exist.") + return None, False + + except Exception as exc: + log.warning("Failed to create the alarm: %s", exc) + return None, False + + def alarming(self, message, vim_uuid): """Consume info from the message bus to manage alarms.""" - values = json.loads(message.value) - self.common = common + 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() + auth_token = Common.get_auth_token(vim_uuid) - if self.endpoint is None: - log.info("Generating a new endpoint for Aodh.") - self.endpoint = self.common.get_endpoint("alarming") + alarm_endpoint = Common.get_endpoint("alarming", vim_uuid) + metric_endpoint = Common.get_endpoint("metric", vim_uuid) + + vim_account = self._auth_manager.get_credentials(vim_uuid) + vim_config = json.loads(vim_account.config) if message.key == "create_alarm_request": # Configure/Update an alarm alarm_details = values['alarm_create_request'] alarm_id, alarm_status = self.configure_alarm( - self.endpoint, self.auth_token, alarm_details) + alarm_endpoint, metric_endpoint, auth_token, alarm_details, vim_config) # Generate a valid response message, send via producer + if alarm_status is True: + 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'].lower(), + alarm_details['ns_id'].lower() + ) 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_resonse', resp_message, - 'alarm_response') - except Exception as exc: - log.warn("Response creation failed: %s", exc) + 'create_alarm_response', resp_message) + except Exception: + log.exception("Response creation failed:") elif message.key == "list_alarm_request": - # Check for a specifed: alarm_name, resource_uuid, severity + # Check for a specified: 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_endpoint, auth_token, list_details) try: # Generate and send a list response back @@ -140,17 +166,16 @@ class Alarming(object): 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.") + 'list_alarm_response', resp_message) + except Exception: + log.exception("Failed to send a valid response back.") 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) + alarm_endpoint, auth_token, alarm_id) # Generate and send a response message try: @@ -160,30 +185,29 @@ class Alarming(object): 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) + 'delete_alarm_response', resp_message) + except Exception: + log.exception("Failed to create delete response: ") elif message.key == "acknowledge_alarm": # Acknowledge that an alarm has been dealt with by the SO alarm_id = values['ack_details']['alarm_uuid'] response = self.update_alarm_state( - self.endpoint, self.auth_token, alarm_id) + alarm_endpoint, auth_token, alarm_id) # 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.") + log.warning("Failed to acknowledge/clear the alarm.") 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) + alarm_endpoint, auth_token, alarm_details, vim_config) # Generate a response for an update request try: @@ -193,57 +217,21 @@ class Alarming(object): 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) + 'update_alarm_response', resp_message) + except Exception: + log.exception("Failed to send an update response: ") else: log.debug("Unknown key, no action will be performed") return - def configure_alarm(self, endpoint, auth_token, values): - """Create requested alarm in Aodh.""" - url = "{}/v2/alarms/".format(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 - - # 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 - def delete_alarm(self, endpoint, auth_token, alarm_id): """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( + result = 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) @@ -252,8 +240,8 @@ class Alarming(object): else: return True - except Exception as exc: - log.warn("Failed to delete alarm: %s because %s.", alarm_id, exc) + except Exception: + log.exception("Failed to delete alarm %s :", alarm_id) return False def list_alarms(self, endpoint, auth_token, list_details): @@ -262,19 +250,16 @@ class Alarming(object): 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) + log.warning("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 @@ -288,7 +273,7 @@ class Alarming(object): # Perform the request to get the desired list try: - result = self.common._perform_request( + result = Common.perform_request( url, auth_token, req_type="get") if result is not None: @@ -296,7 +281,7 @@ class Alarming(object): for alarm in json.loads(result.text): rule = alarm['gnocchi_resources_threshold_rule'] if resource == rule['resource_id']: - res_list.append(str(alarm)) + res_list.append(alarm) if not res_list: log.info("No alarms for this resource") return a_list @@ -307,23 +292,23 @@ class Alarming(object): name, sev) for alarm in json.loads(result.text): if name == alarm['name']: - name_list.append(str(alarm)) + name_list.append(alarm) for alarm in json.loads(result.text): if sev == alarm['severity']: - sev_list.append(str(alarm)) + sev_list.append(alarm) 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) 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) a_list = list(set(sev_list).intersection(res_list)) else: log.info("Returning an entire list of alarms.") @@ -343,77 +328,93 @@ class Alarming(object): payload = json.dumps("ok") try: - self.common._perform_request( + 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) + except Exception: + log.exception("Unable to update alarm state: ") return False - def update_alarm(self, endpoint, auth_token, values): + def update_alarm(self, endpoint, auth_token, values, vim_config): """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( + 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 = rule['metric'] + metric_name = [key for key, value in six.iteritems(METRIC_MAPPINGS) if value == rule['metric']][0] except Exception as exc: - log.warn("Failed to retreive existing alarm info: %s.\ - Can only update OSM alarms.", exc) + log.exception("Failed to retrieve existing alarm info. Can only update OSM alarms.") return None, False # 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( + update_alarm = Common.perform_request( url, auth_token, req_type="put", payload=payload) 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 + log.exception("Alarm update could not be performed: ") return None, False 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: + cfg = Config.instance() # Check state and severity - severity = values['severity'].lower() + + 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() + + 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_name, + 'metric': METRIC_MAPPINGS[metric_name], 'resource_id': resource_id, - 'resource_type': 'generic', - 'aggregation_method': STATISTICS[statistic]} + '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, }) + 'gnocchi_resources_threshold_rule': rule, + 'alarm_actions': [cfg.OS_NOTIFIER_URI], }) return payload except KeyError as exc: - log.warn("Alarm is not configured correctly: %s", exc) + log.warning("Alarm is not configured correctly: %s", exc) return None def get_alarm_state(self, endpoint, auth_token, alarm_id): @@ -421,26 +422,38 @@ class Alarming(object): url = "{}/v2/alarms/%s/state".format(endpoint) % alarm_id try: - alarm_state = self.common._perform_request( + alarm_state = 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) + log.warning("Failed to get the state of the alarm:%s", exc) return None - def check_for_metric(self, auth_token, m_name, r_id): + def check_for_metric(self, auth_token, metric_endpoint, m_name, r_id): """Check for the alarm metric.""" try: - endpoint = self.common.get_endpoint("metric") - - url = "{}/v1/metric/".format(endpoint) - metric_list = self.common._perform_request( + url = "{}/v1/metric?sort=name:asc".format(metric_endpoint) + result = Common.perform_request( url, auth_token, req_type="get") - - for metric in json.loads(metric_list.text): + metric_list = [] + metrics_partial = json.loads(result.text) + for metric in metrics_partial: + metric_list.append(metric) + + while len(json.loads(result.text)) > 0: + last_metric_id = metrics_partial[-1]['id'] + url = "{}/v1/metric?sort=name:asc&marker={}".format(metric_endpoint, last_metric_id) + result = Common.perform_request( + url, auth_token, req_type="get") + if len(json.loads(result.text)) > 0: + metrics_partial = json.loads(result.text) + for metric in metrics_partial: + metric_list.append(metric) + metric_id = None + for metric in metric_list: name = metric['name'] resource = metric['resource_id'] - if (name == m_name and resource == r_id): + if name == METRIC_MAPPINGS[m_name] and resource == r_id: metric_id = metric['id'] log.info("The required metric exists, an alarm will be created.") return metric_id