import json
import logging
import random
+import re
import uuid
-from collections import Iterable
+from string import ascii_lowercase
from kafka import KafkaProducer, KafkaConsumer
-from osm_common import dbmongo
+from n2vc.vnf import N2VC
from prometheus_client.core import GaugeMetricFamily
+from osm_mon.common.common_db_client import CommonDbClient
from osm_mon.core.settings import Config
log = logging.getLogger(__name__)
def __init__(self):
cfg = Config.instance()
self.kafka_server = cfg.BROKER_URI
- self.common_db_host = cfg.MONGO_URI.split(':')[0]
- self.common_db_port = cfg.MONGO_URI.split(':')[1]
- self.common_db = dbmongo.DbMongo()
- self.common_db.db_connect({'host': self.common_db_host, 'port': int(self.common_db_port), 'name': 'osm'})
+ self.common_db_client = CommonDbClient()
+ self.n2vc = N2VC(server=cfg.OSMMON_VCA_HOST, secret=cfg.OSMMON_VCA_SECRET)
self.producer = KafkaProducer(bootstrap_servers=self.kafka_server,
key_serializer=str.encode,
value_serializer=str.encode)
group_id='mon-collector-' + str(uuid.uuid4()))
self.consumer.subscribe(['metric_response'])
- def collect_metrics(self) -> Iterable:
+ async def collect_metrics(self):
+ """
+ Collects vdu metrics. These can be vim and/or n2vc metrics.
+ It checks for monitoring-params or metrics inside vdu section of vnfd, then collects the metric accordingly.
+ If vim related, it sends a metric read request through Kafka, to be handled by mon-proxy.
+ If n2vc related, it uses the n2vc client to obtain the readings.
+ :return: lists of metrics
+ """
# TODO(diazb): Remove dependencies on prometheus_client
log.debug("collect_metrics")
metrics = {}
- vnfrs = self.common_db.get_list('vnfrs')
- for vnfr in vnfrs:
- nsr_id = vnfr['nsr-id-ref']
- vnfd = self.common_db.get_one('vnfds', {"_id": vnfr['vnfd-id']})
- payloads = self._generate_metric_data_payloads(vnfr, vnfd)
- for payload in payloads:
- cor_id = payload['correlation_id']
- metric_name = payload['metric_name']
- vnf_member_index = payload['vnf_member_index']
- vdu_name = payload['vdu_name']
- self.producer.send(topic='metric_request', key='read_metric_data_request',
- value=json.dumps(payload))
- self.producer.flush()
- for message in self.consumer:
- if message.key == 'read_metric_data_response':
- content = json.loads(message.value)
- if content['correlation_id'] == cor_id:
- if len(content['metrics_data']['metrics_series']):
- metric_reading = content['metrics_data']['metrics_series'][-1]
- if metric_name not in metrics.keys():
- metrics[metric_name] = GaugeMetricFamily(
- metric_name,
+ try:
+ vnfrs = self.common_db_client.get_vnfrs()
+ vca_model_name = 'default'
+ for vnfr in vnfrs:
+ nsr_id = vnfr['nsr-id-ref']
+ vnfd = self.common_db_client.get_vnfd(vnfr['vnfd-id'])
+ for vdur in vnfr['vdur']:
+ # This avoids errors when vdur records have not been completely filled
+ if 'name' not in vdur:
+ continue
+ vdu = next(
+ filter(lambda vdu: vdu['id'] == vdur['vdu-id-ref'], vnfd['vdu'])
+ )
+ vnf_member_index = vnfr['member-vnf-index-ref']
+ vdu_name = vdur['name']
+ if 'monitoring-param' in vdu:
+ for param in vdu['monitoring-param']:
+ metric_name = param['nfvi-metric']
+ payload = await self._generate_read_metric_payload(metric_name, nsr_id, vdu_name,
+ vnf_member_index)
+ self.producer.send(topic='metric_request', key='read_metric_data_request',
+ value=json.dumps(payload))
+ self.producer.flush()
+ for message in self.consumer:
+ if message.key == 'read_metric_data_response':
+ content = json.loads(message.value)
+ if content['correlation_id'] == payload['correlation_id']:
+ if len(content['metrics_data']['metrics_series']):
+ metric_reading = content['metrics_data']['metrics_series'][-1]
+ if metric_name not in metrics.keys():
+ metrics[metric_name] = GaugeMetricFamily(
+ metric_name,
+ 'OSM metric',
+ labels=['ns_id', 'vnf_member_index', 'vdu_name']
+ )
+ metrics[metric_name].add_metric([nsr_id, vnf_member_index, vdu_name],
+ metric_reading)
+ break
+ if 'vdu-configuration' in vdu and 'metrics' in vdu['vdu-configuration']:
+ vnf_name_vca = await self._generate_vca_vdu_name(vdu_name)
+ vnf_metrics = await self.n2vc.GetMetrics(vca_model_name, vnf_name_vca)
+ log.debug('VNF Metrics: %s', vnf_metrics)
+ for vnf_metric_list in vnf_metrics.values():
+ for vnf_metric in vnf_metric_list:
+ log.debug("VNF Metric: %s", vnf_metric)
+ if vnf_metric['key'] not in metrics.keys():
+ metrics[vnf_metric['key']] = GaugeMetricFamily(
+ vnf_metric['key'],
'OSM metric',
labels=['ns_id', 'vnf_member_index', 'vdu_name']
)
- metrics[metric_name].add_metric([nsr_id, vnf_member_index, vdu_name],
- metric_reading)
- break
- return metrics.values()
+ metrics[vnf_metric['key']].add_metric([nsr_id, vnf_member_index, vdu_name],
+ float(vnf_metric['value']))
+ log.debug("metric.values = %s", metrics.values())
+ return metrics.values()
+ except Exception as e:
+ log.exception("Error collecting metrics")
+ raise e
@staticmethod
- def _generate_metric_data_payloads(vnfr: dict, vnfd: dict) -> list:
- log.debug('_generate_metric_data_payloads')
- payloads = []
- nsr_id = vnfr['nsr-id-ref']
- for vdur in vnfr['vdur']:
- # This avoids errors when vdur records have not been completely filled
- if 'name' not in vdur:
- continue
- vdu = next(
- filter(lambda vdu: vdu['id'] == vdur['vdu-id-ref'], vnfd['vdu'])
- )
- if 'monitoring-param' in vdu:
- for param in vdu['monitoring-param']:
- metric_name = param['nfvi-metric']
- vnf_member_index = vnfr['member-vnf-index-ref']
- vdu_name = vdur['name']
- cor_id = random.randint(1, 10e7)
- payload = {
- 'correlation_id': cor_id,
- 'metric_name': metric_name,
- 'ns_id': nsr_id,
- 'vnf_member_index': vnf_member_index,
- 'vdu_name': vdu_name,
- 'collection_period': 1,
- 'collection_unit': 'DAY',
- }
- payloads.append(payload)
- return payloads
+ async def _generate_vca_vdu_name(vdu_name) -> str:
+ """
+ Replaces all digits in vdu name for corresponding ascii characters. This is the format required by N2VC.
+ :param vdu_name: Vdu name according to the vdur
+ :return: Name with digits replaced with characters
+ """
+ vnf_name_vca = ''.join(
+ ascii_lowercase[int(char)] if char.isdigit() else char for char in vdu_name)
+ vnf_name_vca = re.sub(r'-[a-z]+$', '', vnf_name_vca)
+ return vnf_name_vca
+
+ @staticmethod
+ async def _generate_read_metric_payload(metric_name, nsr_id, vdu_name, vnf_member_index) -> dict:
+ """
+ Builds JSON payload for asking for a metric measurement in MON. It follows the model defined in core.models.
+ :param metric_name: OSM metric name (e.g.: cpu_utilization)
+ :param nsr_id: NSR ID
+ :param vdu_name: Vdu name according to the vdur
+ :param vnf_member_index: Index of the VNF in the NS according to the vnfr
+ :return: JSON payload as dict
+ """
+ cor_id = random.randint(1, 10e7)
+ payload = {
+ 'correlation_id': cor_id,
+ 'metric_name': metric_name,
+ 'ns_id': nsr_id,
+ 'vnf_member_index': vnf_member_index,
+ 'vdu_name': vdu_name,
+ 'collection_period': 1,
+ 'collection_unit': 'DAY',
+ }
+ return payload
projects / osm / MON.git / blobdiff