Refactors project document cleanup to use shared registry utility

Simplifies logic for identifying opened projects by introducing a shared utility function.
Improves reliability of project document removal by ensuring active projects are preserved.
Adds comprehensive unit tests to verify correct behavior in various edge cases.

Author: matusdrobuliak66

services/web/server/src/simcore_service_webserver/garbage_collector/_core_orphans.py

@@ -7,7 +7,6 @@
 from models_library.api_schemas_dynamic_scheduler.dynamic_services import (
     DynamicServiceStop,
 )
-from models_library.projects import ProjectID
 from models_library.projects_nodes_io import NodeID
 from servicelib.common_headers import UNDEFINED_DEFAULT_SIMCORE_USER_AGENT_VALUE
 from servicelib.logging_errors import create_troubleshootting_log_kwargs
@@ -21,6 +20,7 @@
 )
 from ..projects.api import has_user_project_access_rights
 from ..resource_manager.registry import RedisResourceRegistry
+from ..resource_manager.registry_utils import list_opened_project_ids
 from ..users import users_service
 from ..users.exceptions import UserNotFoundError
 
@@ -71,16 +71,6 @@ async def _remove_service(
         )
 
 
-async def _list_opened_project_ids(registry: RedisResourceRegistry) -> list[ProjectID]:
-    opened_projects: list[ProjectID] = []
-    all_session_alive, _ = await registry.get_all_resource_keys()
-    for alive_session in all_session_alive:
-        resources = await registry.get_resources(alive_session)
-        if "project_id" in resources:
-            opened_projects.append(ProjectID(resources["project_id"]))
-    return opened_projects
-
-
 async def remove_orphaned_services(
     registry: RedisResourceRegistry, app: web.Application
 ) -> None:
@@ -105,7 +95,7 @@ async def remove_orphaned_services(
         service.node_uuid: service for service in running_services
     }
 
-    known_opened_project_ids = await _list_opened_project_ids(registry)
+    known_opened_project_ids = await list_opened_project_ids(registry)
 
     # NOTE: Always skip orphan repmoval when `list_node_ids_in_project` raises an error.
     # Why? If a service is running but the nodes form the correspondign project cannot be listed,

services/web/server/src/simcore_service_webserver/projects/_project_document_service.py

@@ -25,6 +25,8 @@
     get_redis_document_manager_client_sdk,
     get_redis_lock_manager_client_sdk,
 )
+from ..resource_manager.registry import get_registry
+from ..resource_manager.registry_utils import list_opened_project_ids
 from ..socketio._utils import get_socket_server
 from . import _projects_repository
 
@@ -108,6 +110,11 @@ async def remove_project_documents_as_admin(app: web.Application) -> None:
     # Get socketio server instance
     sio = get_socket_server(app)
 
+    # Get known opened projects ids based on Redis resources table
+    registry = get_registry(app)
+    known_opened_project_ids = await list_opened_project_ids(registry)
+    known_opened_project_ids = set(known_opened_project_ids)
+
     projects_removed = 0
 
     # Scan through all project document keys
@@ -125,7 +132,15 @@ async def remove_project_documents_as_admin(app: web.Application) -> None:
         project_uuid = ProjectID(project_uuid_str)
         project_room = SocketIORoomStr.from_project_id(project_uuid)
 
-        # Check if there are any users connected to this project room
+        # 1. CHECK - Check if the project UUID is in the known opened projects
+        if project_uuid in known_opened_project_ids:
+            _logger.debug(
+                "Project %s is in Redis Resources table (which means Project is opened), keeping document",
+                project_uuid,
+            )
+            continue
+
+        # 2. CHECK - Check if there are any users connected to this project room
         try:
             # Get all session IDs (socket IDs) in the project room
             room_sessions = list(
@@ -141,17 +156,16 @@ async def remove_project_documents_as_admin(app: web.Application) -> None:
                     project_uuid,
                 )
             else:
-                _logger.debug(
-                    "Project %s has %d connected users, keeping document",
+                _logger.error(
+                    "Project %s has %d connected users in the socket io room (This is not expected, as project resource is not in the Redis Resources table), keeping document just in case",
                     project_uuid,
                     len(room_sessions),
                 )
 
-        except (KeyError, AttributeError, ValueError) as e:
-            _logger.warning(
-                "Failed to check room participants for project %s: %s",
+        except (KeyError, AttributeError, ValueError):
+            _logger.exception(
+                "Failed to check room participants for project %s",
                 project_uuid,
-                str(e),
             )
             continue
 

services/web/server/src/simcore_service_webserver/resource_manager/registry_utils.py

@@ -0,0 +1,18 @@
+import logging
+
+from models_library.projects import ProjectID
+
+from .registry import RedisResourceRegistry
+
+_logger = logging.getLogger(__name__)
+
+
+async def list_opened_project_ids(registry: RedisResourceRegistry) -> list[ProjectID]:
+    """Lists all project IDs that are currently opened in active sessions."""
+    opened_projects: list[ProjectID] = []
+    all_session_alive, _ = await registry.get_all_resource_keys()
+    for alive_session in all_session_alive:
+        resources = await registry.get_resources(alive_session)
+        if "project_id" in resources:
+            opened_projects.append(ProjectID(resources["project_id"]))
+    return opened_projects