Revert "Removes POL code from MON repo"
[osm/MON.git] / policy_module / osm_policy_module / core / agent.py
diff --git a/policy_module/osm_policy_module/core/agent.py b/policy_module/osm_policy_module/core/agent.py
new file mode 100644 (file)
index 0000000..cdd5dfc
--- /dev/null
@@ -0,0 +1,157 @@
+# -*- 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 json
+import logging
+from typing import Dict, List
+
+import yaml
+from kafka import KafkaConsumer
+from osm_policy_module.common.alarm_config import AlarmConfig
+from osm_policy_module.common.lcm_client import LcmClient
+from osm_policy_module.common.mon_client import MonClient
+from osm_policy_module.core.config import Config
+from osm_policy_module.core.database import ScalingRecord, ScalingAlarm
+
+log = logging.getLogger(__name__)
+
+
+class PolicyModuleAgent:
+    def run(self):
+        cfg = Config.instance()
+        # Initialize servers
+        kafka_server = '{}:{}'.format(cfg.get('policy_module', 'kafka_server_host'),
+                                      cfg.get('policy_module', 'kafka_server_port'))
+
+        # Initialize Kafka consumer
+        log.info("Connecting to Kafka server at %s", kafka_server)
+        consumer = KafkaConsumer(bootstrap_servers=kafka_server,
+                                 key_deserializer=bytes.decode,
+                                 value_deserializer=bytes.decode,
+                                 group_id="pm-consumer")
+        consumer.subscribe(['lcm_pm', 'alarm_response'])
+
+        for message in consumer:
+            log.info("Message arrived: %s", message)
+            try:
+                if message.key == 'configure_scaling':
+                    try:
+                        content = json.loads(message.value)
+                    except:
+                        content = yaml.safe_load(message.value)
+                    log.info("Creating scaling record in DB")
+                    # TODO: Use transactions: http://docs.peewee-orm.com/en/latest/peewee/transactions.html
+                    scaling_record = ScalingRecord.create(
+                        nsr_id=content['ns_id'],
+                        name=content['scaling_group_descriptor']['name'],
+                        content=json.dumps(content)
+                    )
+                    log.info("Created scaling record in DB : nsr_id=%s, name=%s, content=%s",
+                             scaling_record.nsr_id,
+                             scaling_record.name,
+                             scaling_record.content)
+                    alarm_configs = self._get_alarm_configs(content)
+                    for config in alarm_configs:
+                        mon_client = MonClient()
+                        log.info("Creating alarm record in DB")
+                        alarm_uuid = mon_client.create_alarm(
+                            metric_name=config.metric_name,
+                            ns_id=scaling_record.nsr_id,
+                            vdu_name=config.vdu_name,
+                            vnf_member_index=config.vnf_member_index,
+                            threshold=config.threshold,
+                            operation=config.operation,
+                            statistic=config.statistic
+                        )
+                        ScalingAlarm.create(
+                            alarm_id=alarm_uuid,
+                            action=config.action,
+                            scaling_record=scaling_record
+                        )
+                if message.key == 'notify_alarm':
+                    content = json.loads(message.value)
+                    alarm_id = 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']
+                    ns_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_id, metric_name, operation, threshold, vdu_name, vnf_member_index, ns_id)
+                    try:
+                        alarm = ScalingAlarm.select().where(ScalingAlarm.alarm_id == alarm_id).get()
+                        lcm_client = LcmClient()
+                        log.info("Sending scaling action message for ns: %s", alarm_id)
+                        lcm_client.scale(alarm.scaling_record.nsr_id, alarm.scaling_record.name, alarm.action)
+                    except ScalingAlarm.DoesNotExist:
+                        log.info("There is no action configured for alarm %s.", alarm_id)
+            except Exception:
+                log.exception("Error consuming message: ")
+
+    def _get_alarm_configs(self, message_content: Dict) -> List[AlarmConfig]:
+        scaling_criterias = message_content['scaling_group_descriptor']['scaling_policy']['scaling_criteria']
+        alarm_configs = []
+        for criteria in scaling_criterias:
+            metric_name = ''
+            scale_out_threshold = criteria['scale_out_threshold']
+            scale_in_threshold = criteria['scale_in_threshold']
+            scale_out_operation = criteria['scale_out_relational_operation']
+            scale_in_operation = criteria['scale_in_relational_operation']
+            statistic = criteria['monitoring_param']['aggregation_type']
+            vdu_name = ''
+            vnf_member_index = ''
+            if 'vdu_monitoring_param' in criteria['monitoring_param']:
+                vdu_name = criteria['monitoring_param']['vdu_monitoring_param']['vdu_name']
+                vnf_member_index = criteria['monitoring_param']['vdu_monitoring_param']['vnf_member_index']
+                metric_name = criteria['monitoring_param']['vdu_monitoring_param']['name']
+            if 'vnf_metric' in criteria['monitoring_param']:
+                # TODO vnf_metric
+                continue
+            if 'vdu_metric' in criteria['monitoring_param']:
+                # TODO vdu_metric
+                continue
+            scale_out_alarm_config = AlarmConfig(metric_name,
+                                                 vdu_name,
+                                                 vnf_member_index,
+                                                 scale_out_threshold,
+                                                 scale_out_operation,
+                                                 statistic,
+                                                 'scale_out')
+            scale_in_alarm_config = AlarmConfig(metric_name,
+                                                vdu_name,
+                                                vnf_member_index,
+                                                scale_in_threshold,
+                                                scale_in_operation,
+                                                statistic,
+                                                'scale_in')
+            alarm_configs.append(scale_in_alarm_config)
+            alarm_configs.append(scale_out_alarm_config)
+        return alarm_configs