trs/wip/metrics

lib/id3c/api/__init__.py

@@ -4,6 +4,7 @@
 import logging
 from flask import Flask
 from . import config
+from .metrics import metrics
 from .routes import blueprints
 
 
@@ -17,6 +18,8 @@ def create_app():
     for blueprint in blueprints:
         app.register_blueprint(blueprint)
 
+    metrics.init_app(app)
+
     LOG.debug(f"app root is {app.root_path}")
     LOG.debug(f"app static directory is {app.static_folder}")
 

lib/id3c/api/metrics.py

@@ -0,0 +1,19 @@
+"""
+Web API metrics.
+"""
+import prometheus_flask_exporter
+import prometheus_flask_exporter.multiprocess
+
+
+if "prometheus_multiproc_dir" in os.environ:
+    FlaskMetrics = prometheus_flask_exporter.multiprocess.MultiprocessPrometheusMetrics
+else:
+    FlaskMetrics = prometheus_flask_exporter.PrometheusMetrics
+
+
+# This instance is used by both our routes and create_app().
+metrics = FlaskMetrics(
+    app = None,
+    path = None,
+    defaults_prefix = prometheus_flask_exporter.NO_PREFIX,
+    default_latency_as_histogram = False)

lib/id3c/api/routes.py

@@ -4,8 +4,12 @@
 import json
 import logging
 import pkg_resources
-from flask import Blueprint, request, send_file
+import prometheus_client
+from flask import Blueprint, make_response, request, send_file
+
+from ..metrics import DatabaseCollector
 from . import datastore
+from .metrics import metrics
 from .utils.routes import authenticated_datastore_session_required, content_types_accepted, check_content_length
 
 
@@ -20,6 +24,28 @@
 ]
 
 
+# Metrics exposition endpoint
+@api_v1.route("/metrics", methods = ["GET"])
+@metrics.do_not_track()
+@authenticated_datastore_session_required
+def expose_metrics(*, session):
+    """
+    Exposes metrics for Prometheus.
+
+    Includes metrics collected from the Flask app, as well as the database.
+    """
+    registry = prometheus_client.CollectorRegistry(auto_describe = True)
+
+    # Collect metrics from the app-wide registry, potentially from multiple
+    # server processes via files in prometheus_multiproc_dir.
+    registry.register(metrics.registry)
+
+    # Collect metrics from the database using the authenticated session.
+    registry.register(DatabaseCollector(session))
+
+    return make_response(prometheus_client.make_wsgi_app(registry))
+
+
 @api_v1.route("/", methods = ['GET'])
 @api_unversioned.route("/", methods = ['GET'])
 def index():

lib/id3c/metrics.py

@@ -0,0 +1,56 @@
+"""
+Metrics handling functions.
+"""
+import logging
+from prometheus_client.core import GaugeMetricFamily
+from psycopg2.errors import InsufficientPrivilege
+from .db import DatabaseSession
+
+
+LOG = logging.getLogger(__name__)
+
+
+class DatabaseCollector:
+    """
+    Collects metrics from the database, using an existing *session*.
+    """
+    def __init__(self, session: DatabaseSession):
+        self.session = session
+
+
+    def collect(self):
+        with self.session:
+            yield from self.estimated_row_total()
+
+
+    def estimated_row_total(self):
+        family = GaugeMetricFamily(
+            "id3c_estimated_row_total",
+            "Estimated number of rows in an ID3C database table",
+            labels = ("schema", "table"))
+
+        try:
+            metrics = self.session.fetch_all(
+                """
+                select
+                    ns.nspname          as schema,
+                    c.relname           as table,
+                    c.reltuples::bigint as estimated_row_count
+                from
+                    pg_catalog.pg_class c
+                    join pg_catalog.pg_namespace ns on (c.relnamespace = ns.oid)
+                where
+                    ns.nspname in ('receiving', 'warehouse') and
+                    c.relkind = 'r'
+                order by
+                    schema,
+                    "table"
+                """)
+        except InsufficientPrivilege as error:
+            LOG.error(f"Permission denied when collecting id3c_estimated_row_total metrics: {error}")
+            return
+
+        for metric in metrics:
+            family.add_metric((metric.schema, metric.table), metric.estimated_row_count)
+
+        yield family

setup.py

@@ -71,6 +71,8 @@
         "more-itertools",
         "oauth2client >2.0.0,<4.0.0",
         "pandas >=1.0.1,<2",
+        "prometheus_client",
+        "prometheus_flask_exporter",
         "psycopg2 >=2.8,<3",
         "pyyaml",
         "requests",