Implement Energy Consumption Extractor to gather per-VM energy metrics

caso/extract/__init__.py

@@ -19,9 +19,11 @@
 from caso.extract.openstack import CinderExtractor
 from caso.extract.openstack import NeutronExtractor
 from caso.extract.openstack import NovaExtractor
+from caso.extract.prometheus import EnergyConsumptionExtractor
 
 __all__ = [
     "NovaExtractor",
     "CinderExtractor",
     "NeutronExtractor",
+    "EnergyConsumptionExtractor",
 ]

caso/extract/prometheus.py

@@ -0,0 +1,339 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2014 Spanish National Research Council (CSIC)
+#
+# 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.
+
+"""Module containing the Energy Consumption extractor for energy metrics."""
+
+import uuid
+
+import prometheus_api_client
+from oslo_config import cfg
+from oslo_log import log
+
+from caso.extract.openstack import base
+from caso import record
+
+CONF = cfg.CONF
+
+opts = [
+    cfg.StrOpt(
+        "prometheus_endpoint",
+        default="http://localhost:9090",
+        help="Prometheus server endpoint URL.",
+    ),
+    cfg.StrOpt(
+        "prometheus_metric_name",
+        default="prometheus_value",
+        help="Name of the Prometheus metric to query for energy consumption.",
+    ),
+    cfg.StrOpt(
+        "vm_uuid_label_name",
+        default="uuid",
+        help="Name of the label that matches the VM UUID in Prometheus metrics.",
+    ),
+    cfg.ListOpt(
+        "labels",
+        default=["type_instance:scaph_process_power_microwatts"],
+        help="List of label filters as key:value pairs to filter the Prometheus "
+        "metric (e.g., 'type_instance:scaph_process_power_microwatts'). "
+        "The VM UUID label will be added automatically based on vm_uuid_label_name.",
+    ),
+    cfg.IntOpt(
+        "prometheus_step_seconds",
+        default=30,
+        help="Frequency between samples in the time series (in seconds).",
+    ),
+    cfg.StrOpt(
+        "prometheus_query_range",
+        default="1h",
+        help="Query time range (e.g., '1h', '6h', '24h').",
+    ),
+    cfg.BoolOpt(
+        "prometheus_verify_ssl",
+        default=True,
+        help="Whether to verify SSL when connecting to Prometheus.",
+    ),
+    cfg.FloatOpt(
+        "cpu_normalization_factor",
+        default=1.0,
+        help="CPU normalization factor to apply to energy measurements.",
+    ),
+]
+
+CONF.import_opt("site_name", "caso.extract.base")
+CONF.register_opts(opts, group="prometheus")
+
+LOG = log.getLogger(__name__)
+
+
+class EnergyConsumptionExtractor(base.BaseOpenStackExtractor):
+    """An energy consumption extractor for cASO."""
+
+    def __init__(self, project, vo):
+        """Initialize an energy consumption extractor for a given project."""
+        super(EnergyConsumptionExtractor, self).__init__(project, vo)
+        self.nova = self._get_nova_client()
+        self.flavors = self._get_flavors()
+
+    def _get_flavors(self):
+        """Get flavors for the project."""
+        flavors = {}
+        try:
+            for flavor in self.nova.flavors.list(is_public=None):
+                flavors[flavor.id] = flavor.to_dict()
+        except Exception as e:
+            LOG.warning(f"Could not get flavors: {e}")
+        return flavors
+
+    def _energy_consumed_wh(self, vm_uuid):
+        """Calculate the energy consumed (Wh) for a VM from Prometheus.
+
+        This function queries Prometheus for instantaneous power samples
+        (in microwatts) and calculates the energy consumed in Watt-hours.
+
+        :param vm_uuid: UUID of the VM to query energy for
+        :returns: Energy consumed in Watt-hours (Wh)
+        """
+        prom_url = CONF.prometheus.prometheus_endpoint
+        metric_name = CONF.prometheus.prometheus_metric_name
+        step_seconds = CONF.prometheus.prometheus_step_seconds
+        query_range = CONF.prometheus.prometheus_query_range
+        verify_ssl = CONF.prometheus.prometheus_verify_ssl
+        vm_uuid_label_name = CONF.prometheus.vm_uuid_label_name
+        label_filters = CONF.prometheus.labels
+
+        prom = prometheus_api_client.PrometheusConnect(
+            url=prom_url, disable_ssl=not verify_ssl
+        )
+
+        # factor = step_seconds / 3600 converts µW·s to µWh
+        factor = step_seconds / 3600
+
+        # Build labels dictionary from the list of "key:value" strings
+        labels = {}
+        for label_filter in label_filters:
+            if ":" in label_filter:
+                key, value = label_filter.split(":", 1)
+                labels[key.strip()] = value.strip()
+            else:
+                LOG.warning(
+                    f"Invalid label filter format '{label_filter}', "
+                    "expected 'key:value'. Skipping."
+                )
+
+        # Add the VM UUID label
+        labels[vm_uuid_label_name] = vm_uuid
+
+        # Build label string: {key="value", ...}
+        label_selector = ",".join(f'{k}="{v}"' for k, v in labels.items())
+
+        # Construct the PromQL query
+        query = (
+            f"sum_over_time({metric_name}{{{label_selector}}}[{query_range}]) "
+            f"* {factor} / 1000000"
+        )
+
+        LOG.debug(f"Querying Prometheus for VM {vm_uuid} with query: {query}")
+
+        try:
+            # Run query
+            result = prom.custom_query(query=query)
+
+            if not result:
+                LOG.debug(f"No energy data returned for VM {vm_uuid}")
+                return 0.0
+
+            energy_wh = float(result[0]["value"][1])
+            LOG.debug(f"VM {vm_uuid} consumed {energy_wh:.4f} Wh")
+            return energy_wh
+
+        except (KeyError, IndexError, ValueError) as e:
+            LOG.warning(f"Error parsing Prometheus result for VM {vm_uuid}: {e}")
+            return 0.0
+        except Exception as e:
+            LOG.error(f"Error querying Prometheus for VM {vm_uuid}: {e}")
+            return 0.0
+
+    def _get_servers(self, extract_from):
+        """Get all servers for a given date."""
+        servers = []
+        limit = 200
+        marker = None
+        # Use a marker and iter over results until we do not have more to get
+        while True:
+            aux = self.nova.servers.list(
+                search_opts={
+                    "changes-since": extract_from,
+                    "project_id": self.project_id,
+                    "all_tenants": True,
+                },
+                limit=limit,
+                marker=marker,
+            )
+            servers.extend(aux)
+
+            if len(aux) < limit:
+                break
+            marker = aux[-1].id
+
+        return servers
+
+    def _build_energy_record(self, server, energy_value, extract_from, extract_to):
+        """Build an energy consumption record for a VM.
+
+        :param server: Nova server object
+        :param energy_value: Energy consumption value in Wh (raw from Prometheus)
+        :param extract_from: Start time for extraction period
+        :param extract_to: End time for extraction period
+        :returns: EnergyRecord object
+        """
+        vm_uuid = str(server.id)
+        vm_status = server.status.lower()
+
+        # Get server creation time
+        import dateutil.parser
+
+        created_at = dateutil.parser.parse(server.created)
+
+        # Remove timezone info for comparison (extract_from/to are naive)
+        if created_at.tzinfo is not None:
+            created_at = created_at.replace(tzinfo=None)
+
+        # Calculate start and end times for the period
+        start_time = max(created_at, extract_from)
+        end_time = extract_to
+
+        # Calculate durations in seconds
+        duration = (end_time - start_time).total_seconds()
+        wall_clock_time_s = int(duration)
+
+        # For CPU duration, we need to multiply by vCPUs if available
+        # Get flavor info to get vCPUs
+        cpu_count = 1  # Default
+        try:
+            flavor = self.flavors.get(server.flavor.get("id"))
+            if flavor:
+                cpu_count = flavor.get("vcpus", 1)
+        except Exception:
+            pass
+
+        cpu_duration_s = wall_clock_time_s * cpu_count
+
+        # Calculate suspend duration (0 if running)
+        suspend_duration_s = 0
+        if vm_status in ["suspended", "paused"]:
+            suspend_duration_s = wall_clock_time_s
+
+        # ExecUnitFinished: 0 if running, 1 if stopped/deleted
+        exec_unit_finished = 0 if vm_status in ["active", "running"] else 1
+
+        # Get CPU normalization factor from configuration
+        cpu_normalization_factor = CONF.prometheus.cpu_normalization_factor
+
+        # Apply CPU normalization factor to energy
+        energy_wh = energy_value * cpu_normalization_factor
+
+        # Calculate work: CpuDuration_s / Energy_wh
+        # Avoid division by zero
+        if energy_wh > 0:
+            work = cpu_duration_s / energy_wh
+        else:
+            work = 0.0
+
+        # Calculate efficiency: CpuDuration_s / WallClockTime_s
+        # Avoid division by zero
+        if wall_clock_time_s > 0:
+            efficiency = cpu_duration_s / wall_clock_time_s
+        else:
+            efficiency = 0.0
+
+        r = record.EnergyRecord(
+            exec_unit_id=uuid.UUID(vm_uuid),
+            start_exec_time=start_time.strftime("%Y-%m-%dT%H:%M:%SZ"),
+            end_exec_time=end_time.strftime("%Y-%m-%dT%H:%M:%SZ"),
+            energy_wh=energy_wh,
+            work=work,
+            efficiency=efficiency,
+            wall_clock_time_s=wall_clock_time_s,
+            cpu_duration_s=cpu_duration_s,
+            suspend_duration_s=suspend_duration_s,
+            cpu_normalization_factor=cpu_normalization_factor,
+            exec_unit_finished=exec_unit_finished,
+            status=vm_status,
+            owner=self.vo,
+            site_name=CONF.site_name,
+            cloud_type=self.cloud_type,
+            compute_service=CONF.service_name,
+        )
+
+        return r
+
+    def extract(self, extract_from, extract_to):
+        """Extract energy consumption records from Prometheus.
+
+        This method queries Prometheus for energy consumption metrics
+        for each VM in the project.
+
+        :param extract_from: datetime.datetime object indicating the date to
+                             extract records from
+        :param extract_to: datetime.datetime object indicating the date to
+                           extract records to
+        :returns: A list of energy records
+        """
+        # Remove timezone as Nova doesn't expect it
+        extract_from = extract_from.replace(tzinfo=None)
+        extract_to = extract_to.replace(tzinfo=None)
+
+        records = []
+
+        # Get all servers for the project
+        LOG.debug(f"Getting servers for project {self.project}")
+        servers = self._get_servers(extract_from)
+
+        LOG.info(
+            f"Found {len(servers)} VMs for project {self.project}, "
+            f"querying Prometheus for energy metrics"
+        )
+
+        # Query Prometheus for each server
+        for server in servers:
+            vm_uuid = str(server.id)
+            vm_name = server.name
+
+            LOG.debug(f"Querying energy consumption for VM {vm_name} ({vm_uuid})")
+
+            # Get energy consumption using the new method
+            energy_value = self._energy_consumed_wh(vm_uuid)
+
+            if energy_value <= 0:
+                LOG.debug(
+                    f"No energy consumption data for VM {vm_name} ({vm_uuid}), "
+                    "skipping record creation"
+                )
+                continue
+
+            LOG.debug(
+                f"Creating energy record: {energy_value} Wh "
+                f"for VM {vm_name} ({vm_uuid})"
+            )
+
+            energy_record = self._build_energy_record(
+                server, energy_value, extract_from, extract_to
+            )
+            records.append(energy_record)
+
+        LOG.info(f"Extracted {len(records)} energy records for project {self.project}")
+
+        return records

caso/opts.py

@@ -21,6 +21,7 @@
 import caso.extract.base
 import caso.extract.manager
 import caso.extract.openstack.nova
+import caso.extract.prometheus
 import caso.keystone_client
 import caso.manager
 import caso.messenger.logstash
@@ -43,5 +44,6 @@ def list_opts():
         ("benchmark", caso.extract.openstack.nova.benchmark_opts),
         ("keystone_auth", caso.keystone_client.opts),
         ("logstash", caso.messenger.logstash.opts),
+        ("prometheus", caso.extract.prometheus.opts),
         ("ssm", caso.messenger.ssm.opts),
     ]