reduce cardinality of metrics (#1593)

* remove node uuid and project uuid from metrics labels to reduce memory footprint

Author: sanderegg

packages/service-library/src/servicelib/monitor_services.py

@@ -10,25 +10,23 @@
 
 SERVICE_STARTED_LABELS: List[str] = [
     "user_id",
-    "project_id",
-    "service_uuid",
     "service_key",
     "service_tag",
     "service_type",
 ]
 
 SERVICE_STOPPED_LABELS: List[str] = [
     "user_id",
-    "project_id",
-    "service_uuid",
     "service_key",
     "service_tag",
     "service_type",
     "result",
 ]
 
 
-def add_instrumentation(app: web.Application, reg: CollectorRegistry, app_name: str) -> None:
+def add_instrumentation(
+    app: web.Application, reg: CollectorRegistry, app_name: str
+) -> None:
 
     app[kSERVICE_STARTED] = Counter(
         name="services_started_total",
@@ -63,16 +61,12 @@ def service_started(
     # pylint: disable=too-many-arguments
     app: web.Application,
     user_id: str,
-    project_id: str,
-    service_uuid: str,
     service_key: str,
     service_tag: str,
     service_type: Union[ServiceType, str],
 ) -> None:
     app[kSERVICE_STARTED].labels(
         user_id=user_id,
-        project_id=project_id,
-        service_uuid=service_uuid,
         service_key=service_key,
         service_tag=service_tag,
         service_type=service_type,
@@ -83,17 +77,13 @@ def service_stopped(
     # pylint: disable=too-many-arguments
     app: web.Application,
     user_id: str,
-    project_id: str,
-    service_uuid: str,
     service_key: str,
     service_tag: str,
     service_type: Union[ServiceType, str],
     result: Union[ServiceResult, str],
 ) -> None:
     app[kSERVICE_STOPPED].labels(
         user_id=user_id,
-        project_id=project_id,
-        service_uuid=service_uuid,
         service_key=service_key,
         service_tag=service_tag,
         service_type=service_type,

services/director/src/simcore_service_director/producer.py

@@ -756,13 +756,7 @@ async def start_service(
         node_details = containers_meta_data[0]
         if config.MONITORING_ENABLED:
             service_started(
-                app,
-                user_id,
-                project_id,
-                node_uuid,
-                service_key,
-                service_tag,
-                "DYNAMIC",
+                app, user_id, service_key, service_tag, "DYNAMIC",
             )
         # we return only the info of the main service
         return node_details
@@ -939,8 +933,6 @@ async def stop_service(app: web.Application, node_uuid: str) -> None:
             service_stopped(
                 app,
                 "undefined_user",
-                "undefined project",
-                node_uuid,
                 service_details["service_key"],
                 service_details["service_version"],
                 "DYNAMIC",

services/web/server/src/simcore_service_webserver/computation_subscribe.py

@@ -10,7 +10,7 @@
 from tenacity import retry
 
 from servicelib.application_keys import APP_CONFIG_KEY
-from servicelib.monitor_services import service_started, service_stopped
+from servicelib.monitor_services import service_started, service_stopped, SERVICE_STARTED_LABELS, SERVICE_STOPPED_LABELS
 from servicelib.rabbitmq_utils import RabbitMQRetryPolicyUponInitialization
 from simcore_sdk.config.rabbit import Config as RabbitConfig
 
@@ -82,11 +82,11 @@ async def instrumentation_message_handler(
     data = json.loads(message.body)
     if data["metrics"] == "service_started":
         service_started(
-            app, **{key: value for key, value in data.items() if key != "metrics"}
+            app, **{key: data[key] for key in SERVICE_STARTED_LABELS}
         )
     elif data["metrics"] == "service_stopped":
         service_stopped(
-            app, **{key: value for key, value in data.items() if key != "metrics"}
+            app, **{key: data[key] for key in SERVICE_STOPPED_LABELS}
         )
     await message.ack()