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 asyncio
  25 import datetime
  26 import json
  27 import logging
  28 from json import JSONDecodeError
  29
  30 import yaml
  31 from aiokafka import AIOKafkaConsumer
  32
  33 from osm_policy_module.common.common_db_client import CommonDbClient
  34 from osm_policy_module.common.lcm_client import LcmClient
  35 from osm_policy_module.common.mon_client import MonClient
  36 from osm_policy_module.core import database
  37 from osm_policy_module.core.config import Config
  38 from osm_policy_module.core.database import ScalingGroup, ScalingAlarm, ScalingPolicy, ScalingCriteria, DatabaseManager
  39
  40 log = logging.getLogger(__name__)
  41
  42 ALLOWED_KAFKA_KEYS = ['instantiated', 'scaled', 'terminated', 'notify_alarm']
  43
  44
  45 class PolicyModuleAgent:
  46     def __init__(self, loop=None):
  47         cfg = Config.instance()
  48         if not loop:
  49             loop = asyncio.get_event_loop()
  50         self.loop = loop
  51         self.db_client = CommonDbClient()
  52         self.mon_client = MonClient(loop=self.loop)
  53         self.lcm_client = LcmClient(loop=self.loop)
  54         self.kafka_server = '{}:{}'.format(cfg.OSMPOL_MESSAGE_HOST,
  55                                            cfg.OSMPOL_MESSAGE_PORT)
  56         self.database_manager = DatabaseManager()
  57
  58     def run(self):
  59         self.loop.run_until_complete(self.start())
  60
  61     async def start(self):
  62         consumer = AIOKafkaConsumer(
  63             "ns",
  64             "alarm_response",
  65             loop=self.loop,
  66             bootstrap_servers=self.kafka_server,
  67             group_id="pol-consumer",
  68             key_deserializer=bytes.decode,
  69             value_deserializer=bytes.decode,
  70         )
  71         await consumer.start()
  72         try:
  73             async for msg in consumer:
  74                 log.info("Message arrived: %s", msg)
  75                 await self._process_msg(msg.topic, msg.key, msg.value)
  76         finally:
  77             await consumer.stop()
  78
  79     async def _process_msg(self, topic, key, msg):
  80         log.debug("_process_msg topic=%s key=%s msg=%s", topic, key, msg)
  81         try:
  82             if key in ALLOWED_KAFKA_KEYS:
  83                 try:
  84                     content = json.loads(msg)
  85                 except JSONDecodeError:
  86                     content = yaml.safe_load(msg)
  87
  88                 if key == 'instantiated' or key == 'scaled':
  89                     await self._handle_instantiated_or_scaled(content)
  90
  91                 if key == 'terminated':
  92                     await self._handle_terminated(content)
  93
  94                 if key == 'notify_alarm':
  95                     await self._handle_alarm_notification(content)
  96             else:
  97                 log.debug("Key %s is not in ALLOWED_KAFKA_KEYS", key)
  98         except Exception:
  99             log.exception("Error consuming message: ")
 100
 101     async def _handle_alarm_notification(self, content):
 102         log.debug("_handle_alarm_notification: %s", content)
 103         alarm_uuid = content['notify_details']['alarm_uuid']
 104         metric_name = content['notify_details']['metric_name']
 105         operation = content['notify_details']['operation']
 106         threshold = content['notify_details']['threshold_value']
 107         vdu_name = content['notify_details']['vdu_name']
 108         vnf_member_index = content['notify_details']['vnf_member_index']
 109         nsr_id = content['notify_details']['ns_id']
 110         log.info(
 111             "Received alarm notification for alarm %s, \
 112             metric %s, \
 113             operation %s, \
 114             threshold %s, \
 115             vdu_name %s, \
 116             vnf_member_index %s, \
 117             ns_id %s ",
 118             alarm_uuid, metric_name, operation, threshold, vdu_name, vnf_member_index, nsr_id)
 119         try:
 120             alarm = self.database_manager.get_alarm(alarm_uuid)
 121             delta = datetime.datetime.now() - alarm.scaling_criteria.scaling_policy.last_scale
 122             log.debug("last_scale: %s", alarm.scaling_criteria.scaling_policy.last_scale)
 123             log.debug("now: %s", datetime.datetime.now())
 124             log.debug("delta: %s", delta)
 125             if delta.total_seconds() < alarm.scaling_criteria.scaling_policy.cooldown_time:
 126                 log.info("Time between last scale and now is less than cooldown time. Skipping.")
 127                 return
 128             log.info("Sending scaling action message for ns: %s", nsr_id)
 129             await self.lcm_client.scale(nsr_id,
 130                                         alarm.scaling_criteria.scaling_policy.scaling_group.name,
 131                                         alarm.vnf_member_index,
 132                                         alarm.action)
 133             alarm.scaling_criteria.scaling_policy.last_scale = datetime.datetime.now()
 134             alarm.scaling_criteria.scaling_policy.save()
 135         except ScalingAlarm.DoesNotExist:
 136             log.info("There is no action configured for alarm %s.", alarm_uuid)
 137
 138     async def _handle_instantiated_or_scaled(self, content):
 139         log.debug("_handle_instantiated_or_scaled: %s", content)
 140         nslcmop_id = content['nslcmop_id']
 141         nslcmop = self.db_client.get_nslcmop(nslcmop_id)
 142         if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED':
 143             nsr_id = nslcmop['nsInstanceId']
 144             log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id)
 145             await self._configure_scaling_groups(nsr_id)
 146         else:
 147             log.info(
 148                 "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. "
 149                 "Current state is %s. Skipping...",
 150                 nslcmop['operationState'])
 151
 152     async def _handle_terminated(self, content):
 153         log.debug("_handle_deleted: %s", content)
 154         nsr_id = content['nsr_id']
 155         if content['operationState'] == 'COMPLETED' or content['operationState'] == 'PARTIALLY_COMPLETED':
 156             log.info("Deleting scaling groups and alarms for network service with nsr_id: %s", nsr_id)
 157             await self._delete_scaling_groups(nsr_id)
 158         else:
 159             log.info(
 160                 "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. "
 161                 "Current state is %s. Skipping...",
 162                 content['operationState'])
 163
 164     async def _configure_scaling_groups(self, nsr_id: str):
 165         log.debug("_configure_scaling_groups: %s", nsr_id)
 166         alarms_created = []
 167         with database.db.atomic() as tx:
 168             try:
 169                 vnfrs = self.db_client.get_vnfrs(nsr_id)
 170                 for vnfr in vnfrs:
 171                     log.info("Processing vnfr: %s", vnfr)
 172                     vnfd = self.db_client.get_vnfd(vnfr['vnfd-id'])
 173                     log.info("Looking for vnfd %s", vnfr['vnfd-id'])
 174                     if 'scaling-group-descriptor' not in vnfd:
 175                         continue
 176                     scaling_groups = vnfd['scaling-group-descriptor']
 177                     vnf_monitoring_params = vnfd['monitoring-param']
 178                     for scaling_group in scaling_groups:
 179                         try:
 180                             scaling_group_record = ScalingGroup.select().where(
 181                                 ScalingGroup.nsr_id == nsr_id,
 182                                 ScalingGroup.vnf_member_index == int(vnfr['member-vnf-index-ref']),
 183                                 ScalingGroup.name == scaling_group['name']
 184                             ).get()
 185                             log.info("Found existing scaling group record in DB...")
 186                         except ScalingGroup.DoesNotExist:
 187                             log.info("Creating scaling group record in DB...")
 188                             scaling_group_record = ScalingGroup.create(
 189                                 nsr_id=nsr_id,
 190                                 vnf_member_index=vnfr['member-vnf-index-ref'],
 191                                 name=scaling_group['name'],
 192                                 content=json.dumps(scaling_group)
 193                             )
 194                             log.info(
 195                                 "Created scaling group record in DB : nsr_id=%s, vnf_member_index=%s, name=%s",
 196                                 scaling_group_record.nsr_id,
 197                                 scaling_group_record.vnf_member_index,
 198                                 scaling_group_record.name)
 199                         for scaling_policy in scaling_group['scaling-policy']:
 200                             if scaling_policy['scaling-type'] != 'automatic':
 201                                 continue
 202                             try:
 203                                 scaling_policy_record = ScalingPolicy.select().join(ScalingGroup).where(
 204                                     ScalingPolicy.name == scaling_policy['name'],
 205                                     ScalingGroup.id == scaling_group_record.id
 206                                 ).get()
 207                                 log.info("Found existing scaling policy record in DB...")
 208                             except ScalingPolicy.DoesNotExist:
 209                                 log.info("Creating scaling policy record in DB...")
 210                                 scaling_policy_record = ScalingPolicy.create(
 211                                     nsr_id=nsr_id,
 212                                     name=scaling_policy['name'],
 213                                     cooldown_time=scaling_policy['cooldown-time'],
 214                                     scaling_group=scaling_group_record
 215                                 )
 216                                 log.info("Created scaling policy record in DB : name=%s, scaling_group.name=%s",
 217                                          scaling_policy_record.name,
 218                                          scaling_policy_record.scaling_group.name)
 219
 220                             for scaling_criteria in scaling_policy['scaling-criteria']:
 221                                 try:
 222                                     scaling_criteria_record = ScalingCriteria.select().join(ScalingPolicy).where(
 223                                         ScalingPolicy.id == scaling_policy_record.id,
 224                                         ScalingCriteria.name == scaling_criteria['name']
 225                                     ).get()
 226                                     log.info("Found existing scaling criteria record in DB...")
 227                                 except ScalingCriteria.DoesNotExist:
 228                                     log.info("Creating scaling criteria record in DB...")
 229                                     scaling_criteria_record = ScalingCriteria.create(
 230                                         nsr_id=nsr_id,
 231                                         name=scaling_criteria['name'],
 232                                         scaling_policy=scaling_policy_record
 233                                     )
 234                                     log.info(
 235                                         "Created scaling criteria record in DB : name=%s, scaling_policy.name=%s",
 236                                         scaling_criteria_record.name,
 237                                         scaling_criteria_record.scaling_policy.name)
 238
 239                                 vnf_monitoring_param = next(
 240                                     filter(
 241                                         lambda param: param['id'] == scaling_criteria[
 242                                             'vnf-monitoring-param-ref'
 243                                         ],
 244                                         vnf_monitoring_params)
 245                                 )
 246                                 if 'vdu-monitoring-param' in vnf_monitoring_param:
 247                                     vdurs = list(
 248                                         filter(
 249                                             lambda vdur: vdur['vdu-id-ref'] == vnf_monitoring_param
 250                                             ['vdu-monitoring-param']
 251                                             ['vdu-ref'],
 252                                             vnfr['vdur']
 253                                         )
 254                                     )
 255                                 elif 'vdu-metric' in vnf_monitoring_param:
 256                                     vdurs = list(
 257                                         filter(
 258                                             lambda vdur: vdur['vdu-id-ref'] == vnf_monitoring_param
 259                                             ['vdu-metric']
 260                                             ['vdu-ref'],
 261                                             vnfr['vdur']
 262                                         )
 263                                     )
 264                                 elif 'vnf-metric' in vnf_monitoring_param:
 265                                     log.warning("vnf-metric is not currently supported.")
 266                                     continue
 267                                 else:
 268                                     log.warning(
 269                                         "Scaling criteria is referring to a vnf-monitoring-param that does not "
 270                                         "contain a reference to a vdu or vnf metric.")
 271                                     continue
 272                                 for vdur in vdurs:
 273                                     log.info("Creating alarm for vdur %s ", vdur)
 274                                     try:
 275                                         (ScalingAlarm.select()
 276                                          .join(ScalingCriteria)
 277                                          .join(ScalingPolicy)
 278                                          .join(ScalingGroup)
 279                                          .where(
 280                                             ScalingAlarm.vdu_name == vdur['name'],
 281                                             ScalingCriteria.name == scaling_criteria['name'],
 282                                             ScalingPolicy.name == scaling_policy['name'],
 283                                             ScalingGroup.nsr_id == nsr_id
 284                                         ).get())
 285                                         log.debug("vdu %s already has an alarm configured", vdur['name'])
 286                                         continue
 287                                     except ScalingAlarm.DoesNotExist:
 288                                         pass
 289                                     alarm_uuid = await self.mon_client.create_alarm(
 290                                         metric_name=vnf_monitoring_param['id'],
 291                                         ns_id=nsr_id,
 292                                         vdu_name=vdur['name'],
 293                                         vnf_member_index=vnfr['member-vnf-index-ref'],
 294                                         threshold=scaling_criteria['scale-in-threshold'],
 295                                         operation=scaling_criteria['scale-in-relational-operation'],
 296                                         statistic=vnf_monitoring_param['aggregation-type']
 297                                     )
 298                                     alarm = ScalingAlarm.create(
 299                                         alarm_uuid=alarm_uuid,
 300                                         action='scale_in',
 301                                         vnf_member_index=int(vnfr['member-vnf-index-ref']),
 302                                         vdu_name=vdur['name'],
 303                                         scaling_criteria=scaling_criteria_record
 304                                     )
 305                                     alarms_created.append(alarm)
 306                                     alarm_uuid = await self.mon_client.create_alarm(
 307                                         metric_name=vnf_monitoring_param['id'],
 308                                         ns_id=nsr_id,
 309                                         vdu_name=vdur['name'],
 310                                         vnf_member_index=vnfr['member-vnf-index-ref'],
 311                                         threshold=scaling_criteria['scale-out-threshold'],
 312                                         operation=scaling_criteria['scale-out-relational-operation'],
 313                                         statistic=vnf_monitoring_param['aggregation-type']
 314                                     )
 315                                     alarm = ScalingAlarm.create(
 316                                         alarm_uuid=alarm_uuid,
 317                                         action='scale_out',
 318                                         vnf_member_index=int(vnfr['member-vnf-index-ref']),
 319                                         vdu_name=vdur['name'],
 320                                         scaling_criteria=scaling_criteria_record
 321                                     )
 322                                     alarms_created.append(alarm)
 323
 324             except Exception as e:
 325                 log.exception("Error configuring scaling groups:")
 326                 tx.rollback()
 327                 if len(alarms_created) > 0:
 328                     log.info("Cleaning alarm resources in MON")
 329                     for alarm in alarms_created:
 330                         await self.mon_client.delete_alarm(alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id,
 331                                                            alarm.vnf_member_index,
 332                                                            alarm.vdu_name,
 333                                                            alarm.alarm_uuid)
 334                 raise e
 335
 336     async def _delete_scaling_groups(self, nsr_id: str):
 337         with database.db.atomic() as tx:
 338             try:
 339                 for scaling_group in ScalingGroup.select().where(ScalingGroup.nsr_id == nsr_id):
 340                     for scaling_policy in scaling_group.scaling_policies:
 341                         for scaling_criteria in scaling_policy.scaling_criterias:
 342                             for alarm in scaling_criteria.scaling_alarms:
 343                                 try:
 344                                     await self.mon_client.delete_alarm(
 345                                         alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id,
 346                                         alarm.vnf_member_index,
 347                                         alarm.vdu_name,
 348                                         alarm.alarm_uuid)
 349                                 except ValueError:
 350                                     log.exception("Error deleting alarm in MON %s", alarm.alarm_uuid)
 351                                 alarm.delete_instance()
 352                             scaling_criteria.delete_instance()
 353                         scaling_policy.delete_instance()
 354                     scaling_group.delete_instance()
 355
 356             except Exception as e:
 357                 log.exception("Error deleting scaling groups and alarms:")
 358                 tx.rollback()
 359                 raise e