🐛 Fixes issues with dy-sidecar removal by director-v2 (ITISFoundation#3029)

* extended API to raise errors

* fix service removal

* pylint

* fixed method spec

* downgrading log message

* refactor error handling and propagation

* pylint

* refactor to simplify error handling

* cleanup

* remove unsued

* cleanup

* fix codestyle

Co-authored-by: Andrei Neagu <[email protected]>

Author: GitHK

packages/simcore-sdk/src/simcore_sdk/node_ports_common/exceptions.py

@@ -122,6 +122,7 @@ class NodeNotFound(NodeportsException):
     """The given node_uuid was not found"""
 
     def __init__(self, node_uuid):
+        self.node_uuid = node_uuid
         super().__init__(f"the node id {node_uuid} was not found")
 
 

services/director-v2/src/simcore_service_director_v2/modules/dynamic_sidecar/client_api.py

@@ -17,6 +17,7 @@
 from .errors import (
     DynamicSidecarUnexpectedResponseStatus,
     EntrypointContainerNotFoundError,
+    NodeportsDidNotFindNodeError,
 )
 
 # PC -> SAN improvements to discuss
@@ -246,6 +247,11 @@ async def service_push_output_ports(
         response = await self._client.post(
             url, json=port_keys, timeout=self._save_restore_timeout
         )
+        if response.status_code == status.HTTP_404_NOT_FOUND:
+            json_error = response.json()
+            if json_error.get("code") == "dynamic_sidecar.nodeports.node_not_found":
+                raise NodeportsDidNotFindNodeError(node_uuid=json_error["node_uuid"])
+
         if response.status_code != status.HTTP_204_NO_CONTENT:
             raise DynamicSidecarUnexpectedResponseStatus(response, "output ports push")
 

services/director-v2/src/simcore_service_director_v2/modules/dynamic_sidecar/docker_api.py

@@ -96,7 +96,7 @@ async def docker_client() -> AsyncIterator[aiodocker.docker.Docker]:
     except aiodocker.exceptions.DockerError as e:
         message = "Unexpected error from docker client"
         log_message = f"{message} {e.message}\n{traceback.format_exc()}"
-        log.warning(log_message)
+        log.info(log_message)
         raise GenericDockerError(message, e) from e
     finally:
         if client is not None:

services/director-v2/src/simcore_service_director_v2/modules/dynamic_sidecar/errors.py

@@ -3,6 +3,7 @@
 from aiodocker.exceptions import DockerError
 from httpx import Response
 from models_library.projects_nodes import NodeID
+from pydantic.errors import PydanticErrorMixin
 
 from ...core.errors import DirectorException
 
@@ -52,3 +53,11 @@ def __init__(self, response: Response, msg: Optional[str] = None):
         )
         super().__init__(message)
         self.response = response
+
+
+class NodeportsDidNotFindNodeError(PydanticErrorMixin, DirectorException):
+    code = "dynamic_scheduler.output_ports_pulling.node_not_found"
+    msg_template = (
+        "Could not find node '{node_uuid}' in the database. Did not upload data to S3, "
+        "most likely due to service being removed from the study."
+    )

services/director-v2/src/simcore_service_director_v2/modules/dynamic_sidecar/scheduler/events.py

@@ -27,6 +27,7 @@
 from .._namepsace import get_compose_namespace
 from ..client_api import DynamicSidecarClient, get_dynamic_sidecar_client
 from ..docker_api import (
+    are_all_services_present,
     create_network,
     create_service_and_get_id,
     get_node_id_from_task_for_service,
@@ -49,6 +50,7 @@
     DynamicSidecarUnexpectedResponseStatus,
     EntrypointContainerNotFoundError,
     GenericDockerError,
+    NodeportsDidNotFindNodeError,
 )
 from ..volumes_resolver import DynamicSidecarVolumesPathsResolver
 from .abc import DynamicSchedulerEvent
@@ -472,15 +474,23 @@ async def action(cls, app: FastAPI, scheduler_data: SchedulerData) -> None:
             logger.warning(
                 "Could not contact dynamic-sidecar to begin destruction of %s\n%s",
                 scheduler_data.service_name,
-                str(e),
+                f"{e}",
             )
-
         app_settings: AppSettings = app.state.settings
         dynamic_sidecar_settings: DynamicSidecarSettings = (
             app_settings.DYNAMIC_SERVICES.DYNAMIC_SIDECAR
         )
 
-        if scheduler_data.dynamic_sidecar.service_removal_state.can_save:
+        # only try to save the status if :
+        # - the dynamic-sidecar has finished booting correctly
+        # - it is requested to save the state
+        if (
+            scheduler_data.dynamic_sidecar.service_removal_state.can_save
+            and await are_all_services_present(
+                node_uuid=scheduler_data.node_uuid,
+                dynamic_sidecar_settings=app.state.settings.DYNAMIC_SERVICES.DYNAMIC_SIDECAR,
+            )
+        ):
             dynamic_sidecar_client = get_dynamic_sidecar_client(app)
             dynamic_sidecar_endpoint = scheduler_data.dynamic_sidecar.endpoint
 
@@ -501,7 +511,12 @@ async def action(cls, app: FastAPI, scheduler_data: SchedulerData) -> None:
                             dynamic_sidecar_endpoint,
                         )
                     )
-                await logged_gather(*tasks)
+
+                try:
+                    await logged_gather(*tasks)
+                except NodeportsDidNotFindNodeError as err:
+                    logger.warning("%s", f"{err}")
+
                 logger.info("Ports data pushed by dynamic-sidecar")
             # NOTE: ANE: need to use more specific exception here
             except Exception as e:  # pylint: disable=broad-except
@@ -511,7 +526,7 @@ async def action(cls, app: FastAPI, scheduler_data: SchedulerData) -> None:
                         "state and upload outputs %s\n%s"
                     ),
                     scheduler_data.service_name,
-                    str(e),
+                    f"{e}",
                 )
                 # ensure dynamic-sidecar does not get removed
                 # user data can be manually saved and manual

services/dynamic-sidecar/.env-devel

@@ -22,3 +22,9 @@ REGISTRY_AUTH=false
 REGISTRY_USER=test
 REGISTRY_PW=test
 REGISTRY_SSL=false
+
+S3_ENDPOINT=MINIO
+S3_ACCESS_KEY=mocked
+S3_SECRET_KEY=mocked
+S3_BUCKET_NAME=mocked
+R_CLONE_PROVIDER=MINIO

services/dynamic-sidecar/openapi.json

@@ -510,6 +510,9 @@
           "204": {
             "description": "Successful Response"
           },
+          "404": {
+            "description": "Could not find node_uuid in the database"
+          },
           "422": {
             "description": "Validation Error",
             "content": {

services/dynamic-sidecar/src/simcore_service_dynamic_sidecar/api/containers_extension.py

@@ -195,6 +195,11 @@ async def pull_output_ports(
     summary="Push output ports data",
     response_class=Response,
     status_code=status.HTTP_204_NO_CONTENT,
+    responses={
+        status.HTTP_404_NOT_FOUND: {
+            "description": "Could not find node_uuid in the database"
+        }
+    },
 )
 async def push_output_ports(
     port_keys: Optional[List[str]] = None,

services/dynamic-sidecar/src/simcore_service_dynamic_sidecar/core/application.py

@@ -3,14 +3,15 @@
 
 from fastapi import FastAPI
 from servicelib.fastapi.openapi import override_fastapi_openapi_method
+from simcore_sdk.node_ports_common.exceptions import NodeNotFound
 
 from .._meta import API_VTAG, __version__
 from ..api import main_router
 from ..models.domains.shared_store import SharedStore
 from ..models.schemas.application_health import ApplicationHealth
 from ..modules.directory_watcher import setup_directory_watcher
 from .docker_logs import setup_background_log_fetcher
-from .error_handlers import http_error_handler
+from .error_handlers import http_error_handler, node_not_found_error_handler
 from .errors import BaseDynamicSidecarError
 from .rabbitmq import setup_rabbitmq
 from .remote_debug import setup as remote_debug_setup
@@ -78,6 +79,7 @@ def assemble_application() -> FastAPI:
     application.include_router(main_router)
 
     # error handlers
+    application.add_exception_handler(NodeNotFound, node_not_found_error_handler)
     application.add_exception_handler(BaseDynamicSidecarError, http_error_handler)
 
     # also sets up mounted_volumes

services/dynamic-sidecar/src/simcore_service_dynamic_sidecar/core/error_handlers.py

@@ -1,11 +1,29 @@
+from fastapi import status
 from fastapi.encoders import jsonable_encoder
+from simcore_sdk.node_ports_common.exceptions import NodeNotFound
 from starlette.requests import Request
 from starlette.responses import JSONResponse
 
 from .errors import BaseDynamicSidecarError
 
 
-async def http_error_handler(_: Request, exc: BaseDynamicSidecarError) -> JSONResponse:
+async def http_error_handler(
+    _: Request, exception: BaseDynamicSidecarError
+) -> JSONResponse:
     return JSONResponse(
-        content=jsonable_encoder({"errors": [exc.message]}), status_code=exc.status
+        content=jsonable_encoder({"errors": [exception.message]}),
+        status_code=exception.status,
+    )
+
+
+async def node_not_found_error_handler(
+    _: Request, exception: NodeNotFound
+) -> JSONResponse:
+    error_fields = dict(
+        code="dynamic_sidecar.nodeports.node_not_found",
+        message=f"{exception}",
+        node_uuid=f"{exception.node_uuid}",
+    )
+    return JSONResponse(
+        content=jsonable_encoder(error_fields), status_code=status.HTTP_404_NOT_FOUND
     )