[Requirements] Bump fastapi and uvicorn

dockerfiles/mlrun-api/requirements.txt

@@ -1,4 +1,4 @@
-uvicorn~=0.17.0
+uvicorn~=0.20.0
 dask-kubernetes~=0.11.0
 apscheduler~=3.6
 sqlite3-to-mysql~=1.4

mlrun/api/main.py

@@ -361,7 +361,6 @@ def main():
         "mlrun.api.main:app",
         host="0.0.0.0",
         port=config.httpdb.port,
-        debug=config.httpdb.debug,
         access_log=False,
         timeout_keep_alive=config.httpdb.http_connection_timeout_keep_alive,
     )

mlrun/api/utils/background_tasks.py

@@ -18,7 +18,6 @@
 import typing
 import uuid
 
-import anyio
 import fastapi
 import fastapi.concurrency
 import sqlalchemy.orm
@@ -151,21 +150,11 @@ def get_background_task(
     @mlrun.api.utils.helpers.ensure_running_on_chief
     async def background_task_wrapper(self, name: str, function, *args, **kwargs):
         try:
+            if asyncio.iscoroutinefunction(function):
+                await function(*args, **kwargs)
+            else:
+                await fastapi.concurrency.run_in_threadpool(function, *args, **kwargs)
 
-            # In the current fastapi version, there is a bug in the starlette package it uses for the background tasks.
-            # The bug causes the task to be cancelled if the client's http connection is closed before the task is done.
-            # The bug is fixed in the latest version of starlette & fastapi. We will upgrade in 1.3.0, but until then we
-            # will use this workaround to prevent the task from being cancelled all together.
-            # See https://github.com/encode/starlette/issues/1438
-            # and https://github.com/tiangolo/fastapi/issues/5606
-            # TODO: remove this workaround when upgrading to fastapi 0.87.0
-            with anyio.CancelScope(shield=True):
-                if asyncio.iscoroutinefunction(function):
-                    await function(*args, **kwargs)
-                else:
-                    await fastapi.concurrency.run_in_threadpool(
-                        function, *args, **kwargs
-                    )
         except Exception:
             logger.warning(
                 f"Failed during background task execution: {function.__name__}, exc: {traceback.format_exc()}"

requirements.txt

@@ -6,13 +6,6 @@ GitPython~=3.0
 aiohttp~=3.8
 # 8.1.0+ breaks dask/distributed versions older than 2022.04.0, see here - https://github.com/dask/distributed/pull/6018
 click~=8.0.0
-# fastapi 0.78.0 requires starlette 0.19.1 which requires typing_extensions>=3.10.0
-# kfp 1.8.13 requires typing-extensions>=3.7.4,<5
-# for some reason when installing mlrun on a venv that already had typing-extensions==3.7.4.3 it didn't upgrade it to
-# >=3.10.0 although it was installing starlette 0.19.1
-# therefore adding it explicitly
-typing-extensions>=3.10.0,<5
-# in some environments for some reason the protobuf installed was 3.20.0 instead of 3.20.1,
 # when installing google-cloud-storage which required >=3.20.1, <5 it was upgrading the protobuf version to the latest
 # version and because kfp 1.8.13 requires protobuf>=3.13, <4 it resulted incompatibility between kfp and protobuf
 # this can be removed once kfp will support protobuf > 4
@@ -54,7 +47,7 @@ kubernetes~=12.0
 # TODO: move to API requirements (shouldn't really be here, the sql run db using the API sqldb is preventing us from
 #  separating the SDK and API code) (referring to humanfriendly and fastapi)
 humanfriendly~=8.2
-fastapi~=0.78.0
+fastapi~=0.88.0
 fsspec~=2021.8.1
 v3iofs~=0.1.15
 # 3.4 and above failed builidng in some images - see https://github.com/pyca/cryptography/issues/5771

tests/api/api/test_pipelines.py

@@ -71,9 +71,9 @@ def test_list_pipelines_formats(
     kfp_client_mock: kfp.Client,
 ) -> None:
     for format_ in [
-        mlrun.api.schemas.PipelinesFormat.full,
-        mlrun.api.schemas.PipelinesFormat.metadata_only,
-        mlrun.api.schemas.PipelinesFormat.name_only,
+        mlrun.api.schemas.PipelinesFormat.full.value,
+        mlrun.api.schemas.PipelinesFormat.metadata_only.value,
+        mlrun.api.schemas.PipelinesFormat.name_only.value,
     ]:
         runs = _generate_list_runs_mocks()
         expected_runs = [run.to_dict() for run in runs]
@@ -97,10 +97,10 @@ def test_get_pipeline_formats(
     kfp_client_mock: kfp.Client,
 ) -> None:
     for format_ in [
-        mlrun.api.schemas.PipelinesFormat.full,
-        mlrun.api.schemas.PipelinesFormat.metadata_only,
-        mlrun.api.schemas.PipelinesFormat.summary,
-        mlrun.api.schemas.PipelinesFormat.name_only,
+        mlrun.api.schemas.PipelinesFormat.full.value,
+        mlrun.api.schemas.PipelinesFormat.metadata_only.value,
+        mlrun.api.schemas.PipelinesFormat.summary.value,
+        mlrun.api.schemas.PipelinesFormat.name_only.value,
     ]:
         api_run_detail = _generate_get_run_mock()
         _mock_get_run(kfp_client_mock, api_run_detail)
@@ -119,7 +119,7 @@ def test_get_pipeline_no_project_opa_validation(
     client: fastapi.testclient.TestClient,
     kfp_client_mock: kfp.Client,
 ) -> None:
-    format_ = (mlrun.api.schemas.PipelinesFormat.summary,)
+    format_ = (mlrun.api.schemas.PipelinesFormat.summary.value,)
     project = "project-name"
     mlrun.api.crud.Pipelines().resolve_project_from_pipeline = unittest.mock.Mock(
         return_value=project
@@ -150,10 +150,10 @@ def test_get_pipeline_specific_project(
     kfp_client_mock: kfp.Client,
 ) -> None:
     for format_ in [
-        mlrun.api.schemas.PipelinesFormat.full,
-        mlrun.api.schemas.PipelinesFormat.metadata_only,
-        mlrun.api.schemas.PipelinesFormat.summary,
-        mlrun.api.schemas.PipelinesFormat.name_only,
+        mlrun.api.schemas.PipelinesFormat.full.value,
+        mlrun.api.schemas.PipelinesFormat.metadata_only.value,
+        mlrun.api.schemas.PipelinesFormat.summary.value,
+        mlrun.api.schemas.PipelinesFormat.name_only.value,
     ]:
         project = "project-name"
         api_run_detail = _generate_get_run_mock()
@@ -188,7 +188,7 @@ def test_list_pipelines_specific_project(
     )
     response = client.get(
         f"projects/{project}/pipelines",
-        params={"format": mlrun.api.schemas.PipelinesFormat.name_only},
+        params={"format": mlrun.api.schemas.PipelinesFormat.name_only.value},
     )
     expected_response = mlrun.api.schemas.PipelinesOutput(
         runs=expected_runs, total_size=len(expected_runs), next_page_token=None

tests/api/api/test_projects.py

@@ -15,6 +15,7 @@
 import copy
 import datetime
 import http
+import json.decoder
 import os
 import typing
 import unittest.mock
@@ -25,7 +26,6 @@
 import fastapi.testclient
 import mergedeep
 import pytest
-import simplejson.errors
 import sqlalchemy.orm
 from fastapi.testclient import TestClient
 from sqlalchemy.orm import Session
@@ -123,7 +123,7 @@ def test_redirection_from_worker_to_chief_delete_project(
             assert response.status_code == expected_status
             try:
                 assert response.json() == expected_response
-            except simplejson.errors.JSONDecodeError:
+            except json.decoder.JSONDecodeError:
                 # NO_CONTENT response doesn't return json serializable response
                 assert response.text == expected_response
 
@@ -687,7 +687,7 @@ def test_list_projects_leader_format(
     # list in leader format
     response = client.get(
         "projects",
-        params={"format": mlrun.api.schemas.ProjectsFormat.leader},
+        params={"format": mlrun.api.schemas.ProjectsFormat.leader.value},
         headers={
             mlrun.api.schemas.HeaderNames.projects_role: mlrun.mlconf.httpdb.projects.leader
         },
@@ -736,7 +736,7 @@ def test_projects_crud(
     project_patch = {
         "spec": {
             "description": "lemon",
-            "desired_state": mlrun.api.schemas.ProjectState.archived,
+            "desired_state": mlrun.api.schemas.ProjectState.archived.value,
         }
     }
     response = client.patch(f"projects/{name1}", json=project_patch)
@@ -782,7 +782,7 @@ def test_projects_crud(
 
     # list - full
     response = client.get(
-        "projects", params={"format": mlrun.api.schemas.ProjectsFormat.full}
+        "projects", params={"format": mlrun.api.schemas.ProjectsFormat.full.value}
     )
     projects_output = mlrun.api.schemas.ProjectsOutput(**response.json())
     expected = [project_1, project_2]
@@ -832,7 +832,7 @@ def test_projects_crud(
 
     # list - names only - filter by state
     _list_project_names_and_assert(
-        client, [name1], params={"state": mlrun.api.schemas.ProjectState.archived}
+        client, [name1], params={"state": mlrun.api.schemas.ProjectState.archived.value}
     )
 
     # add function to project 1
@@ -1225,7 +1225,7 @@ def _list_project_names_and_assert(
     client: TestClient, expected_names: typing.List[str], params: typing.Dict = None
 ):
     params = params or {}
-    params["format"] = mlrun.api.schemas.ProjectsFormat.name_only
+    params["format"] = mlrun.api.schemas.ProjectsFormat.name_only.value
     # list - names only - filter by state
     response = client.get(
         "projects",

tests/api/api/test_runs.py

@@ -259,9 +259,9 @@ def test_list_runs_partition_by(db: Session, client: TestClient) -> None:
         client,
         {
             "project": projects[0],
-            "partition-by": mlrun.api.schemas.RunPartitionByField.name,
-            "partition-sort-by": mlrun.api.schemas.SortField.created,
-            "partition-order": mlrun.api.schemas.OrderType.asc,
+            "partition-by": mlrun.api.schemas.RunPartitionByField.name.value,
+            "partition-sort-by": mlrun.api.schemas.SortField.created.value,
+            "partition-order": mlrun.api.schemas.OrderType.asc.value,
         },
         3,
     )
@@ -274,9 +274,9 @@ def test_list_runs_partition_by(db: Session, client: TestClient) -> None:
         client,
         {
             "project": projects[0],
-            "partition-by": mlrun.api.schemas.RunPartitionByField.name,
-            "partition-sort-by": mlrun.api.schemas.SortField.updated,
-            "partition-order": mlrun.api.schemas.OrderType.desc,
+            "partition-by": mlrun.api.schemas.RunPartitionByField.name.value,
+            "partition-sort-by": mlrun.api.schemas.SortField.updated.value,
+            "partition-order": mlrun.api.schemas.OrderType.desc.value,
         },
         3,
     )
@@ -289,9 +289,9 @@ def test_list_runs_partition_by(db: Session, client: TestClient) -> None:
         client,
         {
             "project": projects[0],
-            "partition-by": mlrun.api.schemas.RunPartitionByField.name,
-            "partition-sort-by": mlrun.api.schemas.SortField.updated,
-            "partition-order": mlrun.api.schemas.OrderType.desc,
+            "partition-by": mlrun.api.schemas.RunPartitionByField.name.value,
+            "partition-sort-by": mlrun.api.schemas.SortField.updated.value,
+            "partition-order": mlrun.api.schemas.OrderType.desc.value,
             "rows-per-partition": 5,
         },
         15,
@@ -302,9 +302,9 @@ def test_list_runs_partition_by(db: Session, client: TestClient) -> None:
         client,
         {
             "project": projects[0],
-            "partition-by": mlrun.api.schemas.RunPartitionByField.name,
-            "partition-sort-by": mlrun.api.schemas.SortField.updated,
-            "partition-order": mlrun.api.schemas.OrderType.desc,
+            "partition-by": mlrun.api.schemas.RunPartitionByField.name.value,
+            "partition-sort-by": mlrun.api.schemas.SortField.updated.value,
+            "partition-order": mlrun.api.schemas.OrderType.desc.value,
             "rows-per-partition": 5,
             "max-partitions": 2,
         },
@@ -320,9 +320,9 @@ def test_list_runs_partition_by(db: Session, client: TestClient) -> None:
         {
             "project": projects[0],
             "iter": False,
-            "partition-by": mlrun.api.schemas.RunPartitionByField.name,
-            "partition-sort-by": mlrun.api.schemas.SortField.updated,
-            "partition-order": mlrun.api.schemas.OrderType.desc,
+            "partition-by": mlrun.api.schemas.RunPartitionByField.name.value,
+            "partition-sort-by": mlrun.api.schemas.SortField.updated.value,
+            "partition-order": mlrun.api.schemas.OrderType.desc.value,
             "rows-per-partition": 2,
             "max-partitions": 1,
         },

tests/api/api/test_runtime_resources.py

@@ -83,7 +83,9 @@ def test_list_runtimes_resources_group_by_job(
     )
     response = client.get(
         "projects/*/runtime-resources",
-        params={"group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.job},
+        params={
+            "group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.job.value
+        },
     )
     body = response.json()
     expected_body = {
@@ -201,13 +203,17 @@ def test_list_runtime_resources_no_resources(
     assert body == []
     response = client.get(
         "projects/*/runtime-resources",
-        params={"group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.job},
+        params={
+            "group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.job.value
+        },
     )
     body = response.json()
     assert body == {}
     response = client.get(
         "projects/*/runtime-resources",
-        params={"group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.project},
+        params={
+            "group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.project.value
+        },
     )
     body = response.json()
     assert body == {}
@@ -751,7 +757,9 @@ def _mock_opa_filter_and_assert_list_response(
     )
     response = client.get(
         "projects/*/runtime-resources",
-        params={"group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.project},
+        params={
+            "group-by": mlrun.api.schemas.ListRuntimeResourcesGroupByField.project.value
+        },
     )
     body = response.json()
     expected_body = (

tests/api/api/test_tags.py

@@ -458,7 +458,10 @@ def _delete_artifact_tag(
         ],
         project: str = None,
     ):
-        return client.delete(
+        # using client.request instead of client.delete because the latter doesn't support body
+        # https://www.python-httpx.org/compatibility/#request-body-on-http-methods
+        return client.request(
+            "DELETE",
             API_TAGS_PATH.format(project=project or self.project, tag=tag),
             json=self._generate_tag_identifiers_json(identifiers=identifiers),
         )

tests/api/conftest.py

@@ -68,15 +68,7 @@ def set_base_url_for_test_client(
     client: typing.Union[httpx.AsyncClient, TestClient],
     prefix: str = BASE_VERSIONED_API_PREFIX,
 ):
-    if isinstance(client, httpx.AsyncClient):
-        client.base_url = client.base_url.join(prefix)
-    elif isinstance(client, TestClient):
-        client.base_url = client.base_url + prefix
-
-        # https://stackoverflow.com/questions/10893374/python-confusions-with-urljoin/10893427#10893427
-        client.base_url = client.base_url.rstrip("/") + "/"
-    else:
-        raise NotImplementedError(f"Unknown test client type: {type(client)}")
+    client.base_url = client.base_url.join(prefix)
 
 
 @pytest.fixture()