osm_policy_module/core/agent.py

   1 # -*- coding: utf-8 -*-
   2
   3 # Copyright 2018 Whitestack, LLC
   4 # *************************************************************
   5
   6 # This file is part of OSM Monitoring module
   7 # All Rights Reserved to Whitestack, LLC
   8
   9 # Licensed under the Apache License, Version 2.0 (the "License"); you may
  10 # not use this file except in compliance with the License. You may obtain
  11 # a copy of the License at
  12
  13 #         http://www.apache.org/licenses/LICENSE-2.0
  14
  15 # Unless required by applicable law or agreed to in writing, software
  16 # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
  17 # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
  18 # License for the specific language governing permissions and limitations
  19 # under the License.
  20
  21 # For those usages not covered by the Apache License, Version 2.0 please
  22 # contact: bdiaz@whitestack.com or glavado@whitestack.com
  23 ##
  24 import json
  25 import logging
  26 import threading
  27 from json import JSONDecodeError
  28
  29 import yaml
  30 from kafka import KafkaConsumer
  31
  32 from osm_policy_module.common.db_client import DbClient
  33 from osm_policy_module.common.lcm_client import LcmClient
  34 from osm_policy_module.common.mon_client import MonClient
  35 from osm_policy_module.core import database
  36 from osm_policy_module.core.config import Config
  37 from osm_policy_module.core.database import ScalingGroup, ScalingAlarm, ScalingPolicy, ScalingCriteria
  38
  39 log = logging.getLogger(__name__)
  40
  41 ALLOWED_KAFKA_KEYS = ['instantiated', 'scaled', 'notify_alarm']
  42
  43
  44 class PolicyModuleAgent:
  45     def __init__(self):
  46         cfg = Config.instance()
  47         self.db_client = DbClient()
  48         self.mon_client = MonClient()
  49         self.lcm_client = LcmClient()
  50         self.kafka_server = '{}:{}'.format(cfg.OSMPOL_MESSAGE_HOST,
  51                                            cfg.OSMPOL_MESSAGE_PORT)
  52
  53     def run(self):
  54         consumer = KafkaConsumer(bootstrap_servers=self.kafka_server,
  55                                  key_deserializer=bytes.decode,
  56                                  value_deserializer=bytes.decode,
  57                                  group_id='pol-consumer')
  58         consumer.subscribe(["ns", "alarm_response"])
  59
  60         for message in consumer:
  61             t = threading.Thread(target=self._process_msg, args=(message.topic, message.key, message.value,))
  62             t.start()
  63
  64     def _process_msg(self, topic, key, msg):
  65         try:
  66             log.debug("Message arrived with topic: %s, key: %s, msg: %s", topic, key, msg)
  67             if key in ALLOWED_KAFKA_KEYS:
  68                 try:
  69                     content = json.loads(msg)
  70                 except JSONDecodeError:
  71                     content = yaml.safe_load(msg)
  72
  73                 if key == 'instantiated' or key == 'scaled':
  74                     self._handle_instantiated_or_scaled(content)
  75
  76                 if key == 'notify_alarm':
  77                     self._handle_alarm_notification(content)
  78             else:
  79                 log.debug("Key %s is not in ALLOWED_KAFKA_KEYS", key)
  80         except Exception:
  81             log.exception("Error consuming message: ")
  82
  83     def _handle_alarm_notification(self, content):
  84         alarm_id = content['notify_details']['alarm_uuid']
  85         metric_name = content['notify_details']['metric_name']
  86         operation = content['notify_details']['operation']
  87         threshold = content['notify_details']['threshold_value']
  88         vdu_name = content['notify_details']['vdu_name']
  89         vnf_member_index = content['notify_details']['vnf_member_index']
  90         ns_id = content['notify_details']['ns_id']
  91         log.info(
  92             "Received alarm notification for alarm %s, \
  93             metric %s, \
  94             operation %s, \
  95             threshold %s, \
  96             vdu_name %s, \
  97             vnf_member_index %s, \
  98             ns_id %s ",
  99             alarm_id, metric_name, operation, threshold, vdu_name, vnf_member_index, ns_id)
 100         try:
 101             alarm = ScalingAlarm.select().where(ScalingAlarm.alarm_id == alarm_id).get()
 102             log.info("Sending scaling action message for ns: %s", alarm_id)
 103             self.lcm_client.scale(alarm.scaling_record.nsr_id, alarm.scaling_record.name, alarm.vnf_member_index,
 104                                   alarm.action)
 105         except ScalingAlarm.DoesNotExist:
 106             log.info("There is no action configured for alarm %s.", alarm_id)
 107
 108     def _handle_instantiated_or_scaled(self, content):
 109         nslcmop_id = content['nslcmop_id']
 110         nslcmop = self.db_client.get_nslcmop(nslcmop_id)
 111         if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED':
 112             nsr_id = nslcmop['nsInstanceId']
 113             log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id)
 114             self._configure_scaling_groups(nsr_id)
 115         else:
 116             log.info(
 117                 "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. "
 118                 "Current state is %s. Skipping...",
 119                 nslcmop['operationState'])
 120
 121     def _configure_scaling_groups(self, nsr_id: str):
 122         # TODO(diazb): Check for alarm creation on exception and clean resources if needed.
 123         with database.db.atomic():
 124             vnfrs = self.db_client.get_vnfrs(nsr_id)
 125             log.info("Checking %s vnfrs...", len(vnfrs))
 126             for vnfr in vnfrs:
 127                 vnfd = self.db_client.get_vnfd(vnfr['vnfd-id'])
 128                 log.info("Looking for vnfd %s", vnfr['vnfd-id'])
 129                 scaling_groups = vnfd['scaling-group-descriptor']
 130                 vnf_monitoring_params = vnfd['monitoring-param']
 131                 for scaling_group in scaling_groups:
 132                     try:
 133                         scaling_group_record = ScalingGroup.select().where(
 134                             ScalingGroup.nsr_id == nsr_id,
 135                             ScalingGroup.name == scaling_group['name']
 136                         ).get()
 137                     except ScalingGroup.DoesNotExist:
 138                         log.info("Creating scaling group record in DB...")
 139                         scaling_group_record = ScalingGroup.create(
 140                             nsr_id=nsr_id,
 141                             name=scaling_group['name'],
 142                             content=json.dumps(scaling_group)
 143                         )
 144                         log.info("Created scaling group record in DB : nsr_id=%s, name=%s, content=%s",
 145                                  scaling_group_record.nsr_id,
 146                                  scaling_group_record.name,
 147                                  scaling_group_record.content)
 148                     for scaling_policy in scaling_group['scaling-policy']:
 149                         if scaling_policy['scaling-type'] != 'automatic':
 150                             continue
 151                         try:
 152                             scaling_policy_record = ScalingPolicy.select().join(ScalingGroup).where(
 153                                 ScalingPolicy.name == scaling_policy['name'],
 154                                 ScalingGroup.id == scaling_group_record.id
 155                             ).get()
 156                         except ScalingPolicy.DoesNotExist:
 157                             log.info("Creating scaling policy record in DB...")
 158                             scaling_policy_record = ScalingPolicy.create(
 159                                 nsr_id=nsr_id,
 160                                 name=scaling_policy['name'],
 161                                 scaling_group=scaling_group_record
 162                             )
 163                             log.info("Created scaling policy record in DB : name=%s, scaling_group.name=%s",
 164                                      scaling_policy_record.nsr_id,
 165                                      scaling_policy_record.scaling_group.name)
 166                         for vdu in vnfd['vdu']:
 167                             vdu_monitoring_params = vdu['monitoring-param']
 168                             for scaling_criteria in scaling_policy['scaling-criteria']:
 169                                 try:
 170                                     scaling_criteria_record = ScalingCriteria.select().join(ScalingPolicy).where(
 171                                         ScalingPolicy.id == scaling_policy_record.id,
 172                                         ScalingCriteria.name == scaling_criteria['name']
 173                                     ).get()
 174                                 except ScalingCriteria.DoesNotExist:
 175                                     log.info("Creating scaling criteria record in DB...")
 176                                     scaling_criteria_record = ScalingCriteria.create(
 177                                         nsr_id=nsr_id,
 178                                         name=scaling_policy['name'],
 179                                         scaling_policy=scaling_policy_record
 180                                     )
 181                                     log.info(
 182                                         "Created scaling criteria record in DB : name=%s, scaling_criteria.name=%s",
 183                                         scaling_criteria_record.name,
 184                                         scaling_criteria_record.scaling_policy.name)
 185                                 vnf_monitoring_param = next(
 186                                     filter(lambda param: param['id'] == scaling_criteria['vnf-monitoring-param-ref'],
 187                                            vnf_monitoring_params))
 188                                 # TODO: Add support for non-nfvi metrics
 189                                 vdu_monitoring_param = next(
 190                                     filter(
 191                                         lambda param: param['id'] == vnf_monitoring_param['vdu-monitoring-param-ref'],
 192                                         vdu_monitoring_params))
 193                                 vdurs = list(filter(lambda vdur: vdur['vdu-id-ref'] == vnf_monitoring_param['vdu-ref'],
 194                                                     vnfr['vdur']))
 195                                 for vdur in vdurs:
 196                                     try:
 197                                         ScalingAlarm.select().where(
 198                                             ScalingAlarm.vdu_name == vdur['name']
 199                                         ).where(
 200                                             ScalingAlarm.scaling_criteria.name == scaling_criteria['name']
 201                                         ).get()
 202                                         log.debug("VDU %s already has an alarm configured")
 203                                         continue
 204                                     except ScalingAlarm.DoesNotExist:
 205                                         pass
 206                                     alarm_uuid = self.mon_client.create_alarm(
 207                                         metric_name=vdu_monitoring_param['nfvi-metric'],
 208                                         ns_id=nsr_id,
 209                                         vdu_name=vdur['name'],
 210                                         vnf_member_index=vnfr['member-vnf-index-ref'],
 211                                         threshold=scaling_criteria['scale-in-threshold'],
 212                                         operation=scaling_criteria['scale-in-relational-operation'],
 213                                         statistic=vnf_monitoring_param['aggregation-type']
 214                                     )
 215                                     ScalingAlarm.create(
 216                                         alarm_id=alarm_uuid,
 217                                         action='scale_in',
 218                                         vnf_member_index=int(vnfr['member-vnf-index-ref']),
 219                                         vdu_name=vdur['name'],
 220                                         scaling_criteria=scaling_criteria_record
 221                                     )
 222                                     alarm_uuid = self.mon_client.create_alarm(
 223                                         metric_name=vdu_monitoring_param['nfvi-metric'],
 224                                         ns_id=nsr_id,
 225                                         vdu_name=vdur['name'],
 226                                         vnf_member_index=vnfr['member-vnf-index-ref'],
 227                                         threshold=scaling_criteria['scale-out-threshold'],
 228                                         operation=scaling_criteria['scale-out-relational-operation'],
 229                                         statistic=vnf_monitoring_param['aggregation-type']
 230                                     )
 231                                     ScalingAlarm.create(
 232                                         alarm_id=alarm_uuid,
 233                                         action='scale_out',
 234                                         vnf_member_index=int(vnfr['member-vnf-index-ref']),
 235                                         vdu_name=vdur['name'],
 236                                         scaling_criteria=scaling_criteria_record
 237                                     )