osm_policy_module/autoscaling/service.py

# -*- coding: utf-8 -*-
# pylint: disable=no-member

# Copyright 2018 Whitestack, LLC
# *************************************************************

# This file is part of OSM Monitoring module
# All Rights Reserved to Whitestack, LLC

# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at

#         http://www.apache.org/licenses/LICENSE-2.0

# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.

# For those usages not covered by the Apache License, Version 2.0 please
# contact: bdiaz@whitestack.com or glavado@whitestack.com
##
import asyncio
import datetime
import json
import logging
import operator
import functools

from osm_policy_module.common.common_db_client import CommonDbClient
from osm_policy_module.common.lcm_client import LcmClient
from osm_policy_module.common.mon_client import MonClient
from osm_policy_module.core import database
from osm_policy_module.core.config import Config
from osm_policy_module.core.database import (
    ScalingGroup,
    ScalingAlarm,
    ScalingPolicy,
    ScalingCriteria,
    ScalingAlarmRepository,
    ScalingGroupRepository,
    ScalingPolicyRepository,
    ScalingCriteriaRepository,
)
from osm_policy_module.core.exceptions import VdurNotFound

log = logging.getLogger(__name__)


class AutoscalingService:
    def __init__(self, config: Config, loop=None):
        self.conf = config
        if not loop:
            loop = asyncio.get_event_loop()
        self.loop = loop
        self.db_client = CommonDbClient(config)
        self.mon_client = MonClient(config, loop=self.loop)
        self.lcm_client = LcmClient(config, loop=self.loop)

    async def configure_scaling_groups(self, nsr_id: str, vnf_member_index=None):
        """
        Configures scaling groups for a network service. Creates records in DB. Creates alarms in MON.
        :param nsr_id: Network service record id
        :return:
        """
        log.info(
            "Configuring scaling groups for network service with nsr_id: %s", nsr_id
        )
        alarms_created = []
        database.db.connect()
        try:
            with database.db.atomic() as tx:
                try:
                    if vnf_member_index is None:
                        vnfrs = self.db_client.get_vnfrs(nsr_id)
                    else:
                        vnfrs = []
                        vnfr = self.db_client.get_vnfr(nsr_id, vnf_member_index)
                        vnfrs.append(vnfr)
                    # vnfrs = self.db_client.get_vnfrs(nsr_id)
                    for vnfr in vnfrs:
                        log.debug("Processing vnfr: %s", vnfr)
                        vnfd = self.db_client.get_vnfd(vnfr["vnfd-id"])
                        # TODO: Change for multiple DF support
                        df = vnfd.get("df", [{}])[0]
                        if "scaling-aspect" not in df:
                            log.debug("No scaling aspect present in vnfd")
                            continue
                        # TODO: Change for multiple instantiation levels support
                        scaling_aspects = df["scaling-aspect"]
                        all_vnfd_monitoring_params = (
                            self._get_all_vnfd_monitoring_params(vnfd)
                        )
                        for scaling_aspect in scaling_aspects:
                            scaling_group_record = self._get_or_create_scaling_group(
                                nsr_id, vnfr["member-vnf-index-ref"], scaling_aspect
                            )
                            vdurs = self._get_monitored_vdurs(
                                scaling_aspect, vnfr["vdur"]
                            )
                            for scaling_policy in scaling_aspect.get(
                                "scaling-policy", ()
                            ):
                                if scaling_policy["scaling-type"] != "automatic":
                                    continue
                                scaling_policy_record = (
                                    self._get_or_create_scaling_policy(
                                        nsr_id, scaling_policy, scaling_group_record
                                    )
                                )

                                for scaling_criteria in scaling_policy[
                                    "scaling-criteria"
                                ]:
                                    scaling_criteria_record = (
                                        self._get_or_create_scaling_criteria(
                                            nsr_id,
                                            scaling_criteria,
                                            scaling_policy_record,
                                        )
                                    )
                                    monitoring_param_ref = scaling_criteria.get(
                                        "vnf-monitoring-param-ref"
                                    )
                                    vnf_monitoring_param = all_vnfd_monitoring_params[
                                        monitoring_param_ref
                                    ]

                                    for vdur in vdurs:
                                        vdu_id = vdur["vdu-id-ref"]
                                        log.debug("Creating alarm for vdur %s ", vdur)
                                        try:
                                            ScalingAlarmRepository.get(
                                                ScalingAlarm.vdu_name == vdur["name"],
                                                ScalingCriteria.name
                                                == scaling_criteria["name"],
                                                ScalingPolicy.name
                                                == scaling_policy["name"],
                                                ScalingGroup.nsr_id == nsr_id,
                                                join_classes=[
                                                    ScalingCriteria,
                                                    ScalingPolicy,
                                                    ScalingGroup,
                                                ],
                                            )
                                            log.debug(
                                                "vdu %s already has an alarm configured",
                                                vdur["name"],
                                            )
                                            continue
                                        except ScalingAlarm.DoesNotExist:
                                            pass
                                        metric_name = self._get_metric_name(
                                            vnf_monitoring_param
                                        )

                                        db_nsr = self.db_client.get_nsr(nsr_id)
                                        nb_scale_op = 0
                                        if db_nsr["_admin"].get("scaling-group"):
                                            db_nsr_admin = db_nsr["_admin"][
                                                "scaling-group"
                                            ]
                                            for (
                                                admin_scale_index,
                                                admin_scale_info,
                                            ) in enumerate(db_nsr_admin):
                                                if (
                                                    admin_scale_info["name"]
                                                    == scaling_aspect["name"]
                                                ):
                                                    nb_scale_op = admin_scale_info.get(
                                                        "nb-scale-op", 0
                                                    )
                                                    break
                                        min_instance_count = 1
                                        for vdu_profile in df.get("vdu-profile", ()):
                                            if vdu_profile.get("id") == vdu_id:
                                                min_instance_count = int(
                                                    vdu_profile.get(
                                                        "min-number-of-instances ", 1
                                                    )
                                                )
                                                break
                                        if nb_scale_op >= min_instance_count:
                                            alarm_uuid = (
                                                await self.mon_client.create_alarm(
                                                    metric_name=metric_name,
                                                    ns_id=nsr_id,
                                                    vdu_name=vdur["name"],
                                                    vnf_member_index=vnfr[
                                                        "member-vnf-index-ref"
                                                    ],
                                                    threshold=scaling_criteria[
                                                        "scale-in-threshold"
                                                    ],
                                                    operation=scaling_criteria[
                                                        "scale-in-relational-operation"
                                                    ],
                                                    action="scale_in",
                                                )
                                            )
                                            alarm = ScalingAlarmRepository.create(
                                                alarm_uuid=alarm_uuid,
                                                action="scale_in",
                                                vnf_member_index=vnfr[
                                                    "member-vnf-index-ref"
                                                ],
                                                vdu_name=vdur["name"],
                                                scaling_criteria=scaling_criteria_record,
                                            )
                                            alarms_created.append(alarm)
                                        alarm_uuid = await self.mon_client.create_alarm(
                                            metric_name=metric_name,
                                            ns_id=nsr_id,
                                            vdu_name=vdur["name"],
                                            vnf_member_index=vnfr[
                                                "member-vnf-index-ref"
                                            ],
                                            threshold=scaling_criteria[
                                                "scale-out-threshold"
                                            ],
                                            operation=scaling_criteria[
                                                "scale-out-relational-operation"
                                            ],
                                            action="scale_out",
                                        )
                                        alarm = ScalingAlarmRepository.create(
                                            alarm_uuid=alarm_uuid,
                                            action="scale_out",
                                            vnf_member_index=vnfr[
                                                "member-vnf-index-ref"
                                            ],
                                            vdu_name=vdur["name"],
                                            scaling_criteria=scaling_criteria_record,
                                        )
                                        alarms_created.append(alarm)

                except Exception as e:
                    log.exception("Error configuring scaling groups:")
                    tx.rollback()
                    if len(alarms_created) > 0:
                        log.info("Cleaning alarm resources in MON")
                        for alarm in alarms_created:
                            await self.mon_client.delete_alarm(
                                alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id,
                                alarm.vnf_member_index,
                                alarm.vdu_name,
                                alarm.alarm_uuid,
                            )
                    raise e
        finally:
            database.db.close()

    async def delete_scaling_groups(self, nsr_id: str, vnf_member_index=None):
        log.debug("Deleting scaling groups for network service %s", nsr_id)
        database.db.connect()
        try:
            with database.db.atomic() as tx:
                try:
                    if vnf_member_index is None:
                        scale_conditions = ScalingGroup.nsr_id == nsr_id
                    else:
                        query_list = [
                            ScalingGroup.nsr_id == nsr_id,
                            ScalingGroup.vnf_member_index == vnf_member_index,
                        ]
                        scale_conditions = functools.reduce(operator.and_, query_list)
                    for scaling_group in ScalingGroupRepository.list(scale_conditions):
                        for scaling_policy in scaling_group.scaling_policies:
                            for scaling_criteria in scaling_policy.scaling_criterias:
                                for alarm in scaling_criteria.scaling_alarms:
                                    try:
                                        await self.mon_client.delete_alarm(
                                            alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id,
                                            alarm.vnf_member_index,
                                            alarm.vdu_name,
                                            alarm.alarm_uuid,
                                        )
                                    except ValueError:
                                        log.exception(
                                            "Error deleting alarm in MON %s",
                                            alarm.alarm_uuid,
                                        )
                                    alarm.delete_instance()
                                scaling_criteria.delete_instance()
                            scaling_policy.delete_instance()
                        scaling_group.delete_instance()

                except Exception as e:
                    log.exception("Error deleting scaling groups and alarms:")
                    tx.rollback()
                    raise e
        finally:
            database.db.close()

    async def delete_orphaned_alarms(self, nsr_id):
        log.info("Deleting orphaned scaling alarms for network service %s", nsr_id)
        database.db.connect()
        try:
            with database.db.atomic() as tx:
                try:
                    for scaling_group in ScalingGroupRepository.list(
                        ScalingGroup.nsr_id == nsr_id
                    ):
                        for scaling_policy in scaling_group.scaling_policies:
                            for scaling_criteria in scaling_policy.scaling_criterias:
                                for alarm in scaling_criteria.scaling_alarms:
                                    try:
                                        self.db_client.get_vdur(
                                            nsr_id,
                                            alarm.vnf_member_index,
                                            alarm.vdu_name,
                                        )
                                    except VdurNotFound:
                                        log.debug(
                                            "Deleting orphaned scaling alarm %s",
                                            alarm.alarm_uuid,
                                        )
                                        try:
                                            await self.mon_client.delete_alarm(
                                                alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id,
                                                alarm.vnf_member_index,
                                                alarm.vdu_name,
                                                alarm.alarm_uuid,
                                            )
                                        except ValueError:
                                            log.exception(
                                                "Error deleting alarm in MON %s",
                                                alarm.alarm_uuid,
                                            )
                                        alarm.delete_instance()

                except Exception as e:
                    log.exception("Error deleting orphaned alarms:")
                    tx.rollback()
                    raise e
        finally:
            database.db.close()

    async def handle_alarm(self, alarm_uuid: str, status: str):
        await self.update_alarm_status(alarm_uuid, status)
        await self.evaluate_policy(alarm_uuid)

    async def update_alarm_status(self, alarm_uuid: str, status: str):
        database.db.connect()
        try:
            with database.db.atomic():
                alarm = ScalingAlarmRepository.get(
                    ScalingAlarm.alarm_uuid == alarm_uuid
                )
                alarm.last_status = status
                alarm.save()
        except ScalingAlarm.DoesNotExist:
            log.debug(
                "There is no autoscaling action configured for alarm %s.", alarm_uuid
            )
        finally:
            database.db.close()

    async def evaluate_policy(self, alarm_uuid):
        database.db.connect()
        try:
            with database.db.atomic():
                alarm = ScalingAlarmRepository.get(
                    ScalingAlarm.alarm_uuid == alarm_uuid
                )
                vnf_member_index = alarm.vnf_member_index
                action = alarm.action
                scaling_policy = alarm.scaling_criteria.scaling_policy
                if not scaling_policy.enabled:
                    return
                if action == "scale_in":
                    operation = scaling_policy.scale_in_operation
                elif action == "scale_out":
                    operation = scaling_policy.scale_out_operation
                else:
                    raise Exception("Unknown alarm action {}".format(alarm.action))
                alarms = ScalingAlarmRepository.list(
                    ScalingAlarm.scaling_criteria == alarm.scaling_criteria,
                    ScalingAlarm.action == alarm.action,
                    ScalingAlarm.vnf_member_index == vnf_member_index,
                    ScalingAlarm.vdu_name == alarm.vdu_name,
                )
                statuses = []
                for alarm in alarms:
                    statuses.append(alarm.last_status)
                if (operation == "AND" and set(statuses) == {"alarm"}) or (
                    operation == "OR" and "alarm" in statuses
                ):
                    delta = datetime.datetime.now() - scaling_policy.last_scale
                    if delta.total_seconds() > scaling_policy.cooldown_time:
                        log.info(
                            "Sending %s action message for ns: %s",
                            alarm.action,
                            scaling_policy.scaling_group.nsr_id,
                        )
                        await self.lcm_client.scale(
                            scaling_policy.scaling_group.nsr_id,
                            scaling_policy.scaling_group.name,
                            vnf_member_index,
                            action,
                        )
                        scaling_policy.last_scale = datetime.datetime.now()
                        scaling_policy.save()

        except ScalingAlarm.DoesNotExist:
            log.debug(
                "There is no autoscaling action configured for alarm %s.", alarm_uuid
            )
        finally:
            database.db.close()

    def _get_all_vnfd_monitoring_params(self, vnfd):
        all_monitoring_params = {}
        for ivld in vnfd.get("int-virtual-link-desc", ()):
            for mp in ivld.get("monitoring-parameters", ()):
                all_monitoring_params[mp.get("id")] = mp

        for vdu in vnfd.get("vdu", ()):
            for mp in vdu.get("monitoring-parameter", ()):
                all_monitoring_params[mp.get("id")] = mp

        for df in vnfd.get("df", ()):
            for mp in df.get("monitoring-parameter", ()):
                all_monitoring_params[mp.get("id")] = mp

        return all_monitoring_params

    def _get_or_create_scaling_group(
        self, nsr_id: str, vnf_member_index: str, scaling_aspect: dict
    ):
        try:
            scaling_group_record = ScalingGroupRepository.get(
                ScalingGroup.nsr_id == nsr_id,
                ScalingGroup.vnf_member_index == vnf_member_index,
                ScalingGroup.name == scaling_aspect["name"],
            )
            log.debug("Found existing scaling group record in DB...")
        except ScalingGroup.DoesNotExist:
            log.debug("Creating scaling group record in DB...")
            scaling_group_record = ScalingGroupRepository.create(
                nsr_id=nsr_id,
                vnf_member_index=vnf_member_index,
                name=scaling_aspect["name"],
                content=json.dumps(scaling_aspect),
            )
            log.debug(
                "Created scaling group record in DB : nsr_id=%s, vnf_member_index=%s, name=%s",
                scaling_group_record.nsr_id,
                scaling_group_record.vnf_member_index,
                scaling_group_record.name,
            )
        return scaling_group_record

    def _get_or_create_scaling_policy(
        self, nsr_id: str, scaling_policy: dict, scaling_group_record: ScalingGroup
    ):
        try:
            scaling_policy_record = ScalingPolicyRepository.get(
                ScalingPolicy.name == scaling_policy["name"],
                ScalingGroup.id == scaling_group_record.id,
                join_classes=[ScalingGroup],
            )
            log.debug("Found existing scaling policy record in DB...")
        except ScalingPolicy.DoesNotExist:
            log.debug("Creating scaling policy record in DB...")
            scaling_policy_record = ScalingPolicyRepository.create(
                nsr_id=nsr_id,
                name=scaling_policy["name"],
                cooldown_time=scaling_policy["cooldown-time"],
                scaling_group=scaling_group_record,
            )
            if "scale-in-operation-type" in scaling_policy:
                scaling_policy_record.scale_in_operation = scaling_policy[
                    "scale-in-operation-type"
                ]
            if "scale-out-operation-type" in scaling_policy:
                scaling_policy_record.scale_out_operation = scaling_policy[
                    "scale-out-operation-type"
                ]
            if "enabled" in scaling_policy:
                scaling_policy_record.enabled = scaling_policy["enabled"]
            scaling_policy_record.save()
            log.debug(
                "Created scaling policy record in DB : name=%s, scaling_group.name=%s",
                scaling_policy_record.name,
                scaling_policy_record.scaling_group.name,
            )
        return scaling_policy_record

    def _get_or_create_scaling_criteria(
        self, nsr_id: str, scaling_criteria: dict, scaling_policy_record: ScalingPolicy
    ):
        try:
            scaling_criteria_record = ScalingCriteriaRepository.get(
                ScalingPolicy.id == scaling_policy_record.id,
                ScalingCriteria.name == scaling_criteria["name"],
                join_classes=[ScalingPolicy],
            )
            log.debug("Found existing scaling criteria record in DB...")
        except ScalingCriteria.DoesNotExist:
            log.debug("Creating scaling criteria record in DB...")
            scaling_criteria_record = ScalingCriteriaRepository.create(
                nsr_id=nsr_id,
                name=scaling_criteria["name"],
                scaling_policy=scaling_policy_record,
            )
            log.debug(
                "Created scaling criteria record in DB : name=%s, scaling_policy.name=%s",
                scaling_criteria_record.name,
                scaling_criteria_record.scaling_policy.name,
            )
        return scaling_criteria_record

    def _get_monitored_vdurs(self, scaling_aspect: dict, vdurs):
        all_monitored_vdus = set()
        for delta in scaling_aspect.get("aspect-delta-details", {}).get("deltas", ()):
            for vdu_delta in delta.get("vdu-delta", ()):
                all_monitored_vdus.add(vdu_delta.get("id"))

        monitored_vdurs = list(
            filter(lambda vdur: vdur["vdu-id-ref"] in all_monitored_vdus, vdurs)
        )

        if not monitored_vdurs:
            log.warning(
                "Scaling criteria is referring to a vnf-monitoring-param that does not "
                "contain a reference to a vdu or vnf metric."
            )
        return monitored_vdurs

    def _get_metric_name(self, vnf_monitoring_param: dict):
        if "performance-metric" in vnf_monitoring_param:
            return vnf_monitoring_param["performance-metric"]
        raise ValueError(
            "No metric name found for vnf_monitoring_param %s"
            % vnf_monitoring_param["id"]
        )