mlrun 1.6.0rc21__py3-none-any.whl → 1.6.0rc22__py3-none-any.whl

mlrun/model_monitoring/helpers.py

@@ -28,6 +28,7 @@ from mlrun.utils import logger
 
 if typing.TYPE_CHECKING:
     from mlrun.db.base import RunDBInterface
+    from mlrun.projects import MlrunProject
 
 
 class _BatchDict(typing.TypedDict):
@@ -36,6 +37,10 @@ class _BatchDict(typing.TypedDict):
     days: int
 
 
+class _MLRunNoRunsFoundError(Exception):
+    pass
+
+
 def get_stream_path(project: str = None, application_name: str = None):
     """
     Get stream path from the project secret. If wasn't set, take it from the system configurations
@@ -63,24 +68,22 @@ def get_stream_path(project: str = None, application_name: str = None):
 
 
 def get_monitoring_parquet_path(
-    project: str,
+    project: "MlrunProject",
     kind: str = mlrun.common.schemas.model_monitoring.FileTargetKind.PARQUET,
 ) -> str:
     """Get model monitoring parquet target for the current project and kind. The parquet target path is based on the
     project artifact path. If project artifact path is not defined, the parquet target path will be based on MLRun
     artifact path.
 
-    :param project:     Project name.
+    :param project:     Project object.
     :param kind:        indicate the kind of the parquet path, can be either stream_parquet or stream_controller_parquet
 
     :return:           Monitoring parquet target path.
     """
-
-    project_obj = mlrun.get_or_create_project(name=project)
-    artifact_path = project_obj.spec.artifact_path
+    artifact_path = project.spec.artifact_path
     # Generate monitoring parquet path value
     parquet_path = mlrun.mlconf.get_model_monitoring_file_target_path(
-        project=project,
+        project=project.name,
         kind=kind,
         target="offline",
         artifact_path=artifact_path,
@@ -132,7 +135,7 @@ def _get_monitoring_time_window_from_controller_run(
     run_name = MonitoringFunctionNames.APPLICATION_CONTROLLER
     runs = db.list_runs(project=project, name=run_name, sort=True)
     if not runs:
-        raise MLRunValueError(f"No {run_name} runs were found")
+        raise _MLRunNoRunsFoundError(f"No {run_name} runs were found")
     last_run = runs[0]
     try:
         batch_dict = last_run["spec"]["parameters"]["batch_intervals_dict"]
@@ -162,9 +165,17 @@ def bump_model_endpoint_last_request(
             endpoint_id=model_endpoint.metadata.uid,
         )
         raise MLRunValueError("Model endpoint last request time is empty")
+    try:
+        time_window = _get_monitoring_time_window_from_controller_run(project, db)
+    except _MLRunNoRunsFoundError:
+        logger.debug(
+            "Not bumping model endpoint last request time - no controller runs were found"
+        )
+        return
+
     bumped_last_request = (
         datetime.datetime.fromisoformat(model_endpoint.status.last_request)
-        + _get_monitoring_time_window_from_controller_run(project, db)
+        + time_window
         + datetime.timedelta(
             seconds=mlrun.mlconf.model_endpoint_monitoring.parquet_batching_timeout_secs
         )

mlrun/model_monitoring/stream_processing.py

@@ -528,9 +528,6 @@ class ProcessBeforeTSDB(mlrun.feature_store.steps.MapClass):
 
         # Getting event timestamp and endpoint_id
         base_event = {k: event[k] for k in base_fields}
-        base_event[EventFieldType.TIMESTAMP] = datetime.datetime.fromisoformat(
-            base_event[EventFieldType.TIMESTAMP]
-        )
 
         # base_metrics includes the stats about the average latency and the amount of predictions over time
         base_metrics = {

mlrun/projects/operations.py

@@ -274,7 +274,7 @@ def build_function(
     if not overwrite_build_params:
         # TODO: change overwrite_build_params default to True in 1.8.0
         warnings.warn(
-            "The `overwrite_build_params` parameter default will change from 'False' to 'True in 1.8.0.",
+            "The `overwrite_build_params` parameter default will change from 'False' to 'True' in 1.8.0.",
             mlrun.utils.OverwriteBuildParamsWarning,
         )
 

mlrun/projects/project.py

@@ -2167,7 +2167,7 @@ class MlrunProject(ModelObj):
         self.spec.remove_function(name)
 
     def remove_model_monitoring_function(self, name):
-        """remove the specified model-monitoring-app function from the project
+        """remove the specified model-monitoring-app function from the project and from the db
 
         :param name: name of the model-monitoring-app function (under the project)
         """
@@ -2177,6 +2177,7 @@ class MlrunProject(ModelObj):
             == mm_constants.ModelMonitoringAppLabel.VAL
         ):
             self.remove_function(name=name)
+            mlrun.db.get_run_db().delete_function(name=name.lower())
             logger.info(f"{name} function has been removed from {self.name} project")
         else:
             raise logger.error(
@@ -3016,7 +3017,7 @@ class MlrunProject(ModelObj):
         if not overwrite_build_params:
             # TODO: change overwrite_build_params default to True in 1.8.0
             warnings.warn(
-                "The `overwrite_build_params` parameter default will change from 'False' to 'True in 1.8.0.",
+                "The `overwrite_build_params` parameter default will change from 'False' to 'True' in 1.8.0.",
                 mlrun.utils.OverwriteBuildParamsWarning,
             )
         default_image_name = mlrun.mlconf.default_project_image_name.format(
@@ -3102,7 +3103,7 @@ class MlrunProject(ModelObj):
         if not overwrite_build_params:
             # TODO: change overwrite_build_params default to True in 1.8.0
             warnings.warn(
-                "The `overwrite_build_params` parameter default will change from 'False' to 'True in 1.8.0.",
+                "The `overwrite_build_params` parameter default will change from 'False' to 'True' in 1.8.0.",
                 mlrun.utils.OverwriteBuildParamsWarning,
             )
 
@@ -3407,7 +3408,7 @@ class MlrunProject(ModelObj):
         :param state: List only runs whose state is specified.
         :param sort: Whether to sort the result according to their start time. Otherwise, results will be
             returned by their internal order in the DB (order will not be guaranteed).
-        :param last: Deprecated - currently not used.
+        :param last: Deprecated - currently not used (will be removed in 1.8.0).
         :param iter: If ``True`` return runs from all iterations. Otherwise, return only runs whose ``iter`` is 0.
         :param start_time_from: Filter by run start time in ``[start_time_from, start_time_to]``.
         :param start_time_to: Filter by run start time in ``[start_time_from, start_time_to]``.

mlrun/runtimes/base.py

@@ -550,7 +550,12 @@ class BaseRuntime(ModelObj):
             if err:
                 updates["status.error"] = err_to_str(err)
 
-        elif not was_none and last_state != "completed":
+        elif (
+            not was_none
+            and last_state != mlrun.runtimes.constants.RunStates.completed
+            and last_state
+            not in mlrun.runtimes.constants.RunStates.error_and_abortion_states()
+        ):
             try:
                 runtime_cls = mlrun.runtimes.get_runtime_class(kind)
                 updates = runtime_cls._get_run_completion_updates(resp)

mlrun/runtimes/constants.py

@@ -165,6 +165,17 @@ class RunStates(object):
             RunStates.aborted,
         ]
 
+    @staticmethod
+    def abortion_states():
+        return [
+            RunStates.aborted,
+            RunStates.aborting,
+        ]
+
+    @staticmethod
+    def error_and_abortion_states():
+        return list(set(RunStates.error_states()) | set(RunStates.abortion_states()))
+
     @staticmethod
     def non_terminal_states():
         return list(set(RunStates.all()) - set(RunStates.terminal_states()))

mlrun/runtimes/kubejob.py

@@ -134,7 +134,7 @@ class KubejobRuntime(KubeResource):
         if not overwrite:
             # TODO: change overwrite default to True in 1.8.0
             warnings.warn(
-                "The `overwrite` parameter default will change from 'False' to 'True in 1.8.0.",
+                "The `overwrite` parameter default will change from 'False' to 'True' in 1.8.0.",
                 mlrun.utils.OverwriteBuildParamsWarning,
             )
         image = mlrun.utils.helpers.remove_image_protocol_prefix(image)

mlrun/runtimes/local.py

@@ -307,7 +307,7 @@ class LocalRuntime(BaseRuntime, ParallelRunner):
 
             # if RunError was raised it means that the error was raised as part of running the function
             # ( meaning the state was already updated to error ) therefore we just re-raise the error
-            except (RunError, mlrun.errors.MLRunTaskCancelledError) as err:
+            except RunError as err:
                 raise err
             # this exception handling is for the case where we fail on pre-loading or post-running the function
             # and the state was not updated to error yet, therefore we update the state to error and raise as RunError
@@ -451,58 +451,69 @@ class _DupStdout(object):
 
 def exec_from_params(handler, runobj: RunObject, context: MLClientCtx, cwd=None):
     old_level = logger.level
-    if runobj.spec.verbose:
-        logger.set_logger_level("DEBUG")
-
-    # Prepare the inputs type hints (user may pass type hints as part of the input keys):
-    runobj.spec.extract_type_hints_from_inputs()
-    # Read the keyword arguments to pass to the function (combining params and inputs from the run spec):
-    kwargs = get_func_arg(handler, runobj, context)
-
-    stdout = _DupStdout()
-    err = ""
-    val = None
-    old_dir = os.getcwd()
-    with redirect_stdout(stdout):
-        context.set_logger_stream(stdout)
-        try:
-            if cwd:
-                os.chdir(cwd)
-            # Apply the MLRun handler decorator for parsing inputs using type hints and logging outputs using log hints
-            # (Expected behavior: inputs are being parsed when they have type hints in code or given by user. Outputs
-            # are logged only if log hints are provided by the user):
-            if mlrun.mlconf.packagers.enabled:
-                val = mlrun.handler(
-                    inputs=(
-                        runobj.spec.inputs_type_hints
-                        if runobj.spec.inputs_type_hints
-                        else True  # True will use type hints if provided in user's code.
-                    ),
-                    outputs=(
-                        runobj.spec.returns
-                        if runobj.spec.returns
-                        else None  # None will turn off outputs logging.
-                    ),
-                )(handler)(**kwargs)
-            else:
-                val = handler(**kwargs)
-            context.set_state("completed", commit=False)
-        except Exception as exc:
-            err = err_to_str(exc)
-            logger.error(f"Execution error, {traceback.format_exc()}")
-            context.set_state(error=err, commit=False)
-            logger.set_logger_level(old_level)
-
-    stdout.flush()
-    if cwd:
-        os.chdir(old_dir)
-    context.set_logger_stream(sys.stdout)
-    if val:
-        context.log_result("return", val)
-
-    # completion will be ignored if error is set
-    context.commit(completed=True)
-    logger.set_logger_level(old_level)
+    try:
+        if runobj.spec.verbose:
+            logger.set_logger_level("DEBUG")
+
+        # Prepare the inputs type hints (user may pass type hints as part of the input keys):
+        runobj.spec.extract_type_hints_from_inputs()
+        # Read the keyword arguments to pass to the function (combining params and inputs from the run spec):
+        kwargs = get_func_arg(handler, runobj, context)
+
+        stdout = _DupStdout()
+        err = ""
+        val = None
+        old_dir = os.getcwd()
+        commit = True
+        with redirect_stdout(stdout):
+            context.set_logger_stream(stdout)
+            try:
+                if cwd:
+                    os.chdir(cwd)
+                # Apply the MLRun handler decorator for parsing inputs using type hints and logging outputs using
+                # log hints (Expected behavior: inputs are being parsed when they have type hints in code or given
+                # by user. Outputs are logged only if log hints are provided by the user):
+                if mlrun.mlconf.packagers.enabled:
+                    val = mlrun.handler(
+                        inputs=(
+                            runobj.spec.inputs_type_hints
+                            if runobj.spec.inputs_type_hints
+                            else True  # True will use type hints if provided in user's code.
+                        ),
+                        outputs=(
+                            runobj.spec.returns
+                            if runobj.spec.returns
+                            else None  # None will turn off outputs logging.
+                        ),
+                    )(handler)(**kwargs)
+                else:
+                    val = handler(**kwargs)
+                context.set_state("completed", commit=False)
+            except mlrun.errors.MLRunTaskCancelledError as exc:
+                logger.warning("Run was aborted", err=err_to_str(exc))
+                # Run was aborted, the state run state is updated by the abort job, no need to commit again
+                context.set_state(
+                    mlrun.runtimes.constants.RunStates.aborted, commit=False
+                )
+                commit = False
+            except Exception as exc:
+                err = err_to_str(exc)
+                logger.error(f"Execution error, {traceback.format_exc()}")
+                context.set_state(error=err, commit=False)
+
+        stdout.flush()
+        if cwd:
+            os.chdir(old_dir)
+        context.set_logger_stream(sys.stdout)
+        if val:
+            context.log_result("return", val)
+
+        if commit:
+            # completion will be ignored if error is set
+            context.commit(completed=True)
+
+    finally:
+        logger.set_logger_level(old_level)
     return stdout.buf.getvalue(), err
 
 

mlrun/serving/routers.py

@@ -13,9 +13,11 @@
 # limitations under the License.
 
 import concurrent
+import concurrent.futures
 import copy
 import json
 import traceback
+import typing
 from enum import Enum
 from io import BytesIO
 from typing import Dict, List, Union
@@ -209,16 +211,6 @@ class ModelRouter(BaseModelRouter):
         return event
 
 
-class ExecutorTypes:
-    # TODO: Remove in 1.5.0.
-    thread = "thread"
-    process = "process"
-
-    @staticmethod
-    def all():
-        return [ExecutorTypes.thread, ExecutorTypes.process]
-
-
 class ParallelRunnerModes(str, Enum):
     """Supported parallel running modes for VotingEnsemble"""
 
@@ -313,17 +305,12 @@ class ParallelRun(BaseModelRouter):
         )
         self.name = name or "ParallelRun"
         self.extend_event = extend_event
-        if isinstance(executor_type, ExecutorTypes):
-            executor_type = str(executor_type)
-            logger.warn(
-                "ExecutorTypes is deprecated and will be removed in 1.5.0, use ParallelRunnerModes instead",
-                # TODO: In 1.5.0 to remove ExecutorTypes
-                FutureWarning,
-            )
         self.executor_type = ParallelRunnerModes(executor_type)
-        self._pool: Union[
-            concurrent.futures.ProcessPoolExecutor,
-            concurrent.futures.ThreadPoolExecutor,
+        self._pool: typing.Optional[
+            Union[
+                concurrent.futures.ProcessPoolExecutor,
+                concurrent.futures.ThreadPoolExecutor,
+            ]
         ] = None
 
     def _apply_logic(self, results: dict, event=None):

mlrun/serving/server.py

@@ -40,8 +40,6 @@ from .states import RootFlowStep, RouterStep, get_function, graph_root_setter
 from .utils import (
     event_id_key,
     event_path_key,
-    legacy_event_id_key,
-    legacy_event_path_key,
 )
 
 
@@ -257,18 +255,10 @@ class GraphServer(ModelObj):
         context = context or server_context
         event.content_type = event.content_type or self.default_content_type or ""
         if event.headers:
-            # TODO: remove old event id and path keys in 1.6.0
-            if event_id_key in event.headers or legacy_event_id_key in event.headers:
-                event.id = event.headers.get(event_id_key) or event.headers.get(
-                    legacy_event_id_key
-                )
-            if (
-                event_path_key in event.headers
-                or legacy_event_path_key in event.headers
-            ):
-                event.path = event.headers.get(event_path_key) or event.headers.get(
-                    legacy_event_path_key
-                )
+            if event_id_key in event.headers:
+                event.id = event.headers.get(event_id_key)
+            if event_path_key in event.headers:
+                event.path = event.headers.get(event_path_key)
 
         if isinstance(event.body, (str, bytes)) and (
             not event.content_type or event.content_type in ["json", "application/json"]

mlrun/serving/utils.py

@@ -21,9 +21,6 @@ from mlrun.utils import get_in, update_in
 # more info https://github.com/benoitc/gunicorn/issues/2799, this comment can be removed once old keys are removed
 event_id_key = "MLRUN-EVENT-ID"
 event_path_key = "MLRUN-EVENT-PATH"
-# TODO: remove these keys in 1.6.0
-legacy_event_id_key = "MLRUN_EVENT_ID"
-legacy_event_path_key = "MLRUN_EVENT_PATH"
 
 
 def _extract_input_data(input_path, body):

mlrun/utils/helpers.py

@@ -394,8 +394,11 @@ def get_pretty_types_names(types):
     return types[0].__name__
 
 
-def now_date():
-    return datetime.now(timezone.utc)
+def now_date(tz: timezone = timezone.utc) -> datetime:
+    return datetime.now(tz=tz)
+
+
+datetime_now = now_date
 
 
 def to_date_str(d):

mlrun/utils/logger.py

@@ -131,10 +131,10 @@ class Logger(object):
     def level(self):
         return self._logger.level
 
-    def set_logger_level(self, level: Union[str, int]):
+    def set_logger_level(self, level: Union[str, int]) -> None:
         self._logger.setLevel(level)
 
-    def replace_handler_stream(self, handler_name: str, file: IO[str]):
+    def replace_handler_stream(self, handler_name: str, file: IO[str]) -> None:
         for handler in self._logger.handlers:
             if handler.name == handler_name:
                 handler.stream = file
@@ -193,11 +193,11 @@ def _create_formatter_instance(formatter_kind: FormatterKinds) -> logging.Format
 
 
 def create_logger(
-    level: str = None,
+    level: Optional[str] = None,
     formatter_kind: str = FormatterKinds.HUMAN.name,
     name: str = "mlrun",
-    stream=stdout,
-):
+    stream: IO[str] = stdout,
+) -> Logger:
     level = level or config.log_level or "info"
 
     level = logging.getLevelName(level.upper())

mlrun/utils/version/version.json

@@ -1,4 +1,4 @@
 {
-  "git_commit": "0ac528699a3a7c93f4d4c29e631dd8444b7c86c0",
-  "version": "1.6.0-rc21"
+  "git_commit": "ae3fdb8b9ce822eafb3a810f2a44faf8a5d810ca",
+  "version": "1.6.0-rc22"
 }

{mlrun-1.6.0rc21.dist-info → mlrun-1.6.0rc22.dist-info}/METADATA

@@ -1,11 +1,12 @@
 Metadata-Version: 2.1
 Name: mlrun
-Version: 1.6.0rc21
+Version: 1.6.0rc22
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
 Author-email: yaronh@iguazio.com
 License: Apache License 2.0
+Keywords: mlrun,mlops,data-science,machine-learning,experiment-tracking
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: Apache Software License
@@ -18,6 +19,7 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Topic :: Software Development :: Libraries
+Requires-Python: >=3.9, <3.12
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: urllib3 <1.27,>=1.26.9