From fbb5d6d5882461137bf47c313ead6bb913f15e56 Mon Sep 17 00:00:00 2001 From: Benjamin Diaz Date: Mon, 4 Mar 2019 13:59:53 -0300 Subject: [PATCH] Refactors autoscaling module to separate concerns and allow better control of db connections Change-Id: I57034d34ae2fbb3c4aeda784f9542e643d348460 Signed-off-by: Benjamin Diaz --- osm_policy_module/autoscaling/__init__.py | 0 osm_policy_module/autoscaling/agent.py | 148 ++++++++++++++++ .../{core/agent.py => autoscaling/service.py} | 164 ++++-------------- osm_policy_module/cmd/policy_module_agent.py | 2 +- osm_policy_module/core/database.py | 6 +- .../tests/integration/test_policy_agent.py | 9 +- .../tests/unit/core/test_policy_agent.py | 25 +-- 7 files changed, 208 insertions(+), 146 deletions(-) create mode 100644 osm_policy_module/autoscaling/__init__.py create mode 100644 osm_policy_module/autoscaling/agent.py rename osm_policy_module/{core/agent.py => autoscaling/service.py} (73%) diff --git a/osm_policy_module/autoscaling/__init__.py b/osm_policy_module/autoscaling/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/osm_policy_module/autoscaling/agent.py b/osm_policy_module/autoscaling/agent.py new file mode 100644 index 0000000..bc12dc7 --- /dev/null +++ b/osm_policy_module/autoscaling/agent.py @@ -0,0 +1,148 @@ +# -*- coding: utf-8 -*- + +# 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 logging + +import peewee + +from osm_policy_module.autoscaling.service import Service +from osm_policy_module.common.message_bus_client import MessageBusClient +from osm_policy_module.core.config import Config +from osm_policy_module.core.database import ScalingAlarm + +log = logging.getLogger(__name__) + +ALLOWED_KAFKA_KEYS = ['instantiated', 'scaled', 'terminated', 'notify_alarm'] + + +class PolicyModuleAgent: + def __init__(self, config: Config, loop=None): + self.conf = config + if not loop: + loop = asyncio.get_event_loop() + self.loop = loop + self.msg_bus = MessageBusClient(config) + self.service = Service(config, loop) + + def run(self): + self.loop.run_until_complete(self.start()) + + async def start(self): + topics = [ + "ns", + "alarm_response" + ] + await self.msg_bus.aioread(topics, self._process_msg) + log.critical("Exiting...") + + async def _process_msg(self, topic, key, msg): + log.debug("_process_msg topic=%s key=%s msg=%s", topic, key, msg) + log.info("Message arrived: %s", msg) + try: + if key in ALLOWED_KAFKA_KEYS: + + if key == 'instantiated': + await self._handle_instantiated(msg) + + if key == 'scaled': + await self._handle_scaled(msg) + + if key == 'terminated': + await self._handle_terminated(msg) + + if key == 'notify_alarm': + await self._handle_alarm_notification(msg) + else: + log.debug("Key %s is not in ALLOWED_KAFKA_KEYS", key) + except peewee.PeeweeException: + log.exception("Database error consuming message: ") + raise + except Exception: + log.exception("Error consuming message: ") + + async def _handle_alarm_notification(self, content): + log.debug("_handle_alarm_notification: %s", content) + alarm_uuid = content['notify_details']['alarm_uuid'] + metric_name = content['notify_details']['metric_name'] + operation = content['notify_details']['operation'] + threshold = content['notify_details']['threshold_value'] + vdu_name = content['notify_details']['vdu_name'] + vnf_member_index = content['notify_details']['vnf_member_index'] + nsr_id = content['notify_details']['ns_id'] + log.info( + "Received alarm notification for alarm %s, \ + metric %s, \ + operation %s, \ + threshold %s, \ + vdu_name %s, \ + vnf_member_index %s, \ + ns_id %s ", + alarm_uuid, metric_name, operation, threshold, vdu_name, vnf_member_index, nsr_id) + try: + alarm = self.service.get_alarm(alarm_uuid) + await self.service.scale(alarm) + except ScalingAlarm.DoesNotExist: + log.info("There is no action configured for alarm %s.", alarm_uuid) + + async def _handle_instantiated(self, content): + log.debug("_handle_instantiated: %s", content) + nslcmop_id = content['nslcmop_id'] + nslcmop = self.service.get_nslcmop(nslcmop_id) + if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED': + nsr_id = nslcmop['nsInstanceId'] + log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id) + await self.service.configure_scaling_groups(nsr_id) + else: + log.info( + "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " + "Current state is %s. Skipping...", + nslcmop['operationState']) + + async def _handle_scaled(self, content): + log.debug("_handle_scaled: %s", content) + nslcmop_id = content['nslcmop_id'] + nslcmop = self.service.get_nslcmop(nslcmop_id) + if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED': + nsr_id = nslcmop['nsInstanceId'] + log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id) + await self.service.configure_scaling_groups(nsr_id) + log.info("Checking for orphaned alarms to be deleted for network service with nsr_id: %s", nsr_id) + await self.service.delete_orphaned_alarms(nsr_id) + else: + log.info( + "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " + "Current state is %s. Skipping...", + nslcmop['operationState']) + + async def _handle_terminated(self, content): + log.debug("_handle_deleted: %s", content) + nsr_id = content['nsr_id'] + if content['operationState'] == 'COMPLETED' or content['operationState'] == 'PARTIALLY_COMPLETED': + log.info("Deleting scaling groups and alarms for network service with nsr_id: %s", nsr_id) + await self.service.delete_scaling_groups(nsr_id) + else: + log.info( + "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " + "Current state is %s. Skipping...", + content['operationState']) diff --git a/osm_policy_module/core/agent.py b/osm_policy_module/autoscaling/service.py similarity index 73% rename from osm_policy_module/core/agent.py rename to osm_policy_module/autoscaling/service.py index 65c741f..e1f0704 100644 --- a/osm_policy_module/core/agent.py +++ b/osm_policy_module/autoscaling/service.py @@ -26,24 +26,20 @@ import datetime import json import logging -import peewee - from osm_policy_module.common.common_db_client import CommonDbClient from osm_policy_module.common.lcm_client import LcmClient -from osm_policy_module.common.message_bus_client import MessageBusClient 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, DatabaseManager +from osm_policy_module.core.database import ScalingGroup, ScalingAlarm, ScalingPolicy, ScalingCriteria from osm_policy_module.core.exceptions import VdurNotFound from osm_policy_module.utils.vnfd import VnfdUtils log = logging.getLogger(__name__) -ALLOWED_KAFKA_KEYS = ['instantiated', 'scaled', 'terminated', 'notify_alarm'] +class Service: -class PolicyModuleAgent: def __init__(self, config: Config, loop=None): self.conf = config if not loop: @@ -52,127 +48,11 @@ class PolicyModuleAgent: self.db_client = CommonDbClient(config) self.mon_client = MonClient(config, loop=self.loop) self.lcm_client = LcmClient(config, loop=self.loop) - self.database_manager = DatabaseManager(config) - self.msg_bus = MessageBusClient(config) - - def run(self): - self.loop.run_until_complete(self.start()) - - async def start(self): - topics = [ - "ns", - "alarm_response" - ] - await self.msg_bus.aioread(topics, self._process_msg) - log.critical("Exiting...") - - async def _process_msg(self, topic, key, msg): - log.debug("_process_msg topic=%s key=%s msg=%s", topic, key, msg) - log.info("Message arrived: %s", msg) - try: - if key in ALLOWED_KAFKA_KEYS: - - if key == 'instantiated': - await self._handle_instantiated(msg) - - if key == 'scaled': - await self._handle_scaled(msg) - if key == 'terminated': - await self._handle_terminated(msg) - - if key == 'notify_alarm': - await self._handle_alarm_notification(msg) - else: - log.debug("Key %s is not in ALLOWED_KAFKA_KEYS", key) - except peewee.PeeweeException: - log.exception("Database error consuming message: ") - raise - except Exception: - log.exception("Error consuming message: ") - - async def _handle_alarm_notification(self, content): - log.debug("_handle_alarm_notification: %s", content) - alarm_uuid = content['notify_details']['alarm_uuid'] - metric_name = content['notify_details']['metric_name'] - operation = content['notify_details']['operation'] - threshold = content['notify_details']['threshold_value'] - vdu_name = content['notify_details']['vdu_name'] - vnf_member_index = content['notify_details']['vnf_member_index'] - nsr_id = content['notify_details']['ns_id'] - log.info( - "Received alarm notification for alarm %s, \ - metric %s, \ - operation %s, \ - threshold %s, \ - vdu_name %s, \ - vnf_member_index %s, \ - ns_id %s ", - alarm_uuid, metric_name, operation, threshold, vdu_name, vnf_member_index, nsr_id) - try: - alarm = self.database_manager.get_alarm(alarm_uuid) - delta = datetime.datetime.now() - alarm.scaling_criteria.scaling_policy.last_scale - log.debug("last_scale: %s", alarm.scaling_criteria.scaling_policy.last_scale) - log.debug("now: %s", datetime.datetime.now()) - log.debug("delta: %s", delta) - if delta.total_seconds() < alarm.scaling_criteria.scaling_policy.cooldown_time: - log.info("Time between last scale and now is less than cooldown time. Skipping.") - return - log.info("Sending scaling action message for ns: %s", nsr_id) - await self.lcm_client.scale(nsr_id, - alarm.scaling_criteria.scaling_policy.scaling_group.name, - alarm.vnf_member_index, - alarm.action) - alarm.scaling_criteria.scaling_policy.last_scale = datetime.datetime.now() - alarm.scaling_criteria.scaling_policy.save() - except ScalingAlarm.DoesNotExist: - log.info("There is no action configured for alarm %s.", alarm_uuid) - - async def _handle_instantiated(self, content): - log.debug("_handle_instantiated: %s", content) - nslcmop_id = content['nslcmop_id'] - nslcmop = self.db_client.get_nslcmop(nslcmop_id) - if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED': - nsr_id = nslcmop['nsInstanceId'] - log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id) - await self._configure_scaling_groups(nsr_id) - else: - log.info( - "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " - "Current state is %s. Skipping...", - nslcmop['operationState']) - - async def _handle_scaled(self, content): - log.debug("_handle_scaled: %s", content) - nslcmop_id = content['nslcmop_id'] - nslcmop = self.db_client.get_nslcmop(nslcmop_id) - if nslcmop['operationState'] == 'COMPLETED' or nslcmop['operationState'] == 'PARTIALLY_COMPLETED': - nsr_id = nslcmop['nsInstanceId'] - log.info("Configuring scaling groups for network service with nsr_id: %s", nsr_id) - await self._configure_scaling_groups(nsr_id) - log.info("Checking for orphaned alarms to be deleted for network service with nsr_id: %s", nsr_id) - await self._delete_orphaned_alarms(nsr_id) - else: - log.info( - "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " - "Current state is %s. Skipping...", - nslcmop['operationState']) - - async def _handle_terminated(self, content): - log.debug("_handle_deleted: %s", content) - nsr_id = content['nsr_id'] - if content['operationState'] == 'COMPLETED' or content['operationState'] == 'PARTIALLY_COMPLETED': - log.info("Deleting scaling groups and alarms for network service with nsr_id: %s", nsr_id) - await self._delete_scaling_groups(nsr_id) - else: - log.info( - "Network service is not in COMPLETED or PARTIALLY_COMPLETED state. " - "Current state is %s. Skipping...", - content['operationState']) - - async def _configure_scaling_groups(self, nsr_id: str): + async def configure_scaling_groups(self, nsr_id: str): log.debug("_configure_scaling_groups: %s", nsr_id) alarms_created = [] + database.db.connect() with database.db.atomic() as tx: try: vnfrs = self.db_client.get_vnfrs(nsr_id) @@ -346,8 +226,10 @@ class PolicyModuleAgent: alarm.vdu_name, alarm.alarm_uuid) raise e + database.db.close() - async def _delete_scaling_groups(self, nsr_id: str): + async def delete_scaling_groups(self, nsr_id: str): + database.db.connect() with database.db.atomic() as tx: try: for scaling_group in ScalingGroup.select().where(ScalingGroup.nsr_id == nsr_id): @@ -371,8 +253,10 @@ class PolicyModuleAgent: log.exception("Error deleting scaling groups and alarms:") tx.rollback() raise e + database.db.close() - async def _delete_orphaned_alarms(self, nsr_id): + async def delete_orphaned_alarms(self, nsr_id): + database.db.connect() with database.db.atomic() as tx: try: for scaling_group in ScalingGroup.select().where(ScalingGroup.nsr_id == nsr_id): @@ -397,3 +281,31 @@ class PolicyModuleAgent: log.exception("Error deleting orphaned alarms:") tx.rollback() raise e + database.db.close() + + async def scale(self, alarm): + database.db.connect() + with database.db.atomic(): + delta = datetime.datetime.now() - alarm.scaling_criteria.scaling_policy.last_scale + if delta.total_seconds() > alarm.scaling_criteria.scaling_policy.cooldown_time: + log.info("Sending scaling action message for ns: %s", + alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id) + await self.lcm_client.scale(alarm.scaling_criteria.scaling_policy.scaling_group.nsr_id, + alarm.scaling_criteria.scaling_policy.scaling_group.name, + alarm.vnf_member_index, + alarm.action) + alarm.scaling_criteria.scaling_policy.last_scale = datetime.datetime.now() + alarm.scaling_criteria.scaling_policy.save() + else: + log.info("Time between last scale and now is less than cooldown time. Skipping.") + database.db.close() + + def get_alarm(self, alarm_uuid: str): + database.db.connect() + with database.db.atomic(): + alarm = ScalingAlarm.select().where(ScalingAlarm.alarm_uuid == alarm_uuid).get() + database.db.close() + return alarm + + def get_nslcmop(self, nslcmop_id): + return self.db_client.get_nslcmop(nslcmop_id) diff --git a/osm_policy_module/cmd/policy_module_agent.py b/osm_policy_module/cmd/policy_module_agent.py index af2f602..19e8054 100644 --- a/osm_policy_module/cmd/policy_module_agent.py +++ b/osm_policy_module/cmd/policy_module_agent.py @@ -26,7 +26,7 @@ import asyncio import logging import sys -from osm_policy_module.core.agent import PolicyModuleAgent +from osm_policy_module.autoscaling.agent import PolicyModuleAgent from osm_policy_module.core.config import Config from osm_policy_module.core.database import DatabaseManager diff --git a/osm_policy_module/core/database.py b/osm_policy_module/core/database.py index 3ca2e33..ecaa315 100644 --- a/osm_policy_module/core/database.py +++ b/osm_policy_module/core/database.py @@ -76,10 +76,8 @@ class DatabaseManager: db.initialize(connect(config.get('sql', 'database_uri'))) def create_tables(self) -> None: + db.connect() with db.atomic(): router = Router(db, os.path.dirname(migrations.__file__)) router.run() - - def get_alarm(self, alarm_uuid: str): - with db.atomic(): - return ScalingAlarm.select().where(ScalingAlarm.alarm_uuid == alarm_uuid).get() + db.close() diff --git a/osm_policy_module/tests/integration/test_policy_agent.py b/osm_policy_module/tests/integration/test_policy_agent.py index 7ad9944..4a7efbb 100644 --- a/osm_policy_module/tests/integration/test_policy_agent.py +++ b/osm_policy_module/tests/integration/test_policy_agent.py @@ -23,6 +23,7 @@ ## import asyncio import logging +import os import sys import unittest import uuid @@ -35,7 +36,7 @@ from playhouse.db_url import connect from osm_policy_module.common.common_db_client import CommonDbClient from osm_policy_module.common.mon_client import MonClient from osm_policy_module.core import database -from osm_policy_module.core.agent import PolicyModuleAgent +from osm_policy_module.autoscaling.agent import PolicyModuleAgent from osm_policy_module.core.config import Config from osm_policy_module.core.database import ScalingGroup, ScalingAlarm, ScalingPolicy, ScalingCriteria @@ -426,15 +427,17 @@ MODELS = [ScalingGroup, ScalingPolicy, ScalingCriteria, ScalingAlarm] class PolicyModuleAgentTest(unittest.TestCase): def setUp(self): super() - database.db.initialize(connect('sqlite://')) + database.db.initialize(connect('sqlite:///test_db.sqlite')) database.db.bind(MODELS) database.db.connect() database.db.drop_tables(MODELS) database.db.create_tables(MODELS) + database.db.close() self.loop = asyncio.new_event_loop() def tearDown(self): super() + os.remove('test_db.sqlite') @patch.object(DbMongo, 'db_connect', Mock()) @patch.object(KafkaProducer, '__init__') @@ -459,7 +462,7 @@ class PolicyModuleAgentTest(unittest.TestCase): create_alarm.side_effect = _test_configure_scaling_groups_create_alarm config = Config() agent = PolicyModuleAgent(config, self.loop) - self.loop.run_until_complete(agent._configure_scaling_groups("test_nsr_id")) + self.loop.run_until_complete(agent.service.configure_scaling_groups("test_nsr_id")) create_alarm.assert_any_call(metric_name='cirros_vnf_memory_util', ns_id='test_nsr_id', operation='GT', diff --git a/osm_policy_module/tests/unit/core/test_policy_agent.py b/osm_policy_module/tests/unit/core/test_policy_agent.py index 7fc2dc9..57b0db1 100644 --- a/osm_policy_module/tests/unit/core/test_policy_agent.py +++ b/osm_policy_module/tests/unit/core/test_policy_agent.py @@ -27,20 +27,20 @@ import unittest from unittest import mock from unittest.mock import Mock -from osm_policy_module.core.agent import PolicyModuleAgent +from osm_policy_module.autoscaling.agent import PolicyModuleAgent +from osm_policy_module.autoscaling.service import Service from osm_policy_module.core.config import Config -from osm_policy_module.core.database import DatabaseManager class PolicyAgentTest(unittest.TestCase): def setUp(self): self.loop = asyncio.new_event_loop() - @mock.patch('osm_policy_module.core.agent.CommonDbClient') - @mock.patch('osm_policy_module.core.agent.MonClient') - @mock.patch('osm_policy_module.core.agent.LcmClient') - @mock.patch.object(PolicyModuleAgent, '_configure_scaling_groups') - @mock.patch.object(PolicyModuleAgent, '_delete_orphaned_alarms') + @mock.patch('osm_policy_module.autoscaling.service.CommonDbClient') + @mock.patch('osm_policy_module.autoscaling.service.MonClient') + @mock.patch('osm_policy_module.autoscaling.service.LcmClient') + @mock.patch.object(Service, 'configure_scaling_groups') + @mock.patch.object(Service, 'delete_orphaned_alarms') def test_handle_instantiated(self, delete_orphaned_alarms, configure_scaling_groups, lcm_client, mon_client, db_client): async def mock_configure_scaling_groups(nsr_id): @@ -77,11 +77,12 @@ class PolicyAgentTest(unittest.TestCase): self.loop.run_until_complete(agent._handle_instantiated(content)) configure_scaling_groups.assert_not_called() - @mock.patch('osm_policy_module.core.agent.CommonDbClient') - @mock.patch('osm_policy_module.core.agent.MonClient') - @mock.patch('osm_policy_module.core.agent.LcmClient') - @mock.patch.object(DatabaseManager, 'get_alarm') - def test_handle_alarm_notification(self, get_alarm, lcm_client, mon_client, db_client): + @mock.patch('osm_policy_module.autoscaling.service.CommonDbClient') + @mock.patch('osm_policy_module.autoscaling.service.MonClient') + @mock.patch('osm_policy_module.autoscaling.service.LcmClient') + @mock.patch('osm_policy_module.core.database.db') + @mock.patch.object(Service, 'get_alarm') + def test_handle_alarm_notification(self, get_alarm, db, lcm_client, mon_client, db_client): async def mock_scale(nsr_id, scaling_group_name, vnf_member_index, action): pass -- 2.25.1