mlrun 1.10.0rc8__py3-none-any.whl → 1.10.0rc10__py3-none-any.whl

mlrun/runtimes/mpijob/abstract.py

@@ -54,7 +54,6 @@ class MPIResourceSpec(KubeResourceSpec):
         preemption_mode=None,
         security_context=None,
         state_thresholds=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -84,7 +83,6 @@ class MPIResourceSpec(KubeResourceSpec):
             preemption_mode=preemption_mode,
             security_context=security_context,
             state_thresholds=state_thresholds,
-            serving_spec=serving_spec,
         )
         self.mpi_args = mpi_args or [
             "-x",

mlrun/runtimes/mpijob/v1.py

@@ -49,7 +49,6 @@ class MPIV1ResourceSpec(MPIResourceSpec):
         preemption_mode=None,
         security_context=None,
         state_thresholds=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -80,7 +79,6 @@ class MPIV1ResourceSpec(MPIResourceSpec):
             preemption_mode=preemption_mode,
             security_context=security_context,
             state_thresholds=state_thresholds,
-            serving_spec=serving_spec,
         )
         self.clean_pod_policy = clean_pod_policy or MPIJobV1CleanPodPolicies.default()
 

mlrun/runtimes/nuclio/function.py

@@ -154,7 +154,6 @@ class NuclioSpec(KubeResourceSpec):
         add_templated_ingress_host_mode=None,
         state_thresholds=None,
         disable_default_http_trigger=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -184,7 +183,6 @@ class NuclioSpec(KubeResourceSpec):
             preemption_mode=preemption_mode,
             security_context=security_context,
             state_thresholds=state_thresholds,
-            serving_spec=serving_spec,
         )
 
         self.base_spec = base_spec or {}

mlrun/runtimes/nuclio/serving.py

@@ -42,8 +42,6 @@ from mlrun.serving.states import (
 )
 from mlrun.utils import get_caller_globals, logger, set_paths
 
-from .. import KubejobRuntime
-from ..pod import KubeResourceSpec
 from .function import NuclioSpec, RemoteRuntime, min_nuclio_versions
 
 serving_subkind = "serving_v2"
@@ -151,7 +149,6 @@ class ServingSpec(NuclioSpec):
         state_thresholds=None,
         disable_default_http_trigger=None,
         model_endpoint_creation_task_name=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -192,7 +189,6 @@ class ServingSpec(NuclioSpec):
             service_type=service_type,
             add_templated_ingress_host_mode=add_templated_ingress_host_mode,
             disable_default_http_trigger=disable_default_http_trigger,
-            serving_spec=serving_spec,
         )
 
         self.models = models or {}
@@ -707,7 +703,6 @@ class ServingRuntime(RemoteRuntime):
             "track_models": self.spec.track_models,
             "default_content_type": self.spec.default_content_type,
             "model_endpoint_creation_task_name": self.spec.model_endpoint_creation_task_name,
-            "filename": getattr(self.spec, "filename", None),
         }
 
         if self.spec.secret_sources:
@@ -716,10 +711,6 @@ class ServingRuntime(RemoteRuntime):
 
         return json.dumps(serving_spec)
 
-    @property
-    def serving_spec(self):
-        return self._get_serving_spec()
-
     def to_mock_server(
         self,
         namespace=None,
@@ -824,40 +815,3 @@ class ServingRuntime(RemoteRuntime):
             "Turn off the mock (mock=False) and make sure Nuclio is installed for real deployment to Nuclio"
         )
         self._mock_server = self.to_mock_server()
-
-    def to_job(self) -> KubejobRuntime:
-        """Convert this ServingRuntime to a KubejobRuntime, so that the graph can be run as a standalone job."""
-        if self.spec.function_refs:
-            raise mlrun.errors.MLRunInvalidArgumentError(
-                f"Cannot convert function '{self.metadata.name}' to a job because it has child functions"
-            )
-
-        spec = KubeResourceSpec(
-            image=self.spec.image,
-            mode=self.spec.mode,
-            volumes=self.spec.volumes,
-            volume_mounts=self.spec.volume_mounts,
-            env=self.spec.env,
-            resources=self.spec.resources,
-            default_handler="mlrun.serving.server.execute_graph",
-            pythonpath=self.spec.pythonpath,
-            entry_points=self.spec.entry_points,
-            description=self.spec.description,
-            workdir=self.spec.workdir,
-            image_pull_secret=self.spec.image_pull_secret,
-            node_name=self.spec.node_name,
-            node_selector=self.spec.node_selector,
-            affinity=self.spec.affinity,
-            disable_auto_mount=self.spec.disable_auto_mount,
-            priority_class_name=self.spec.priority_class_name,
-            tolerations=self.spec.tolerations,
-            preemption_mode=self.spec.preemption_mode,
-            security_context=self.spec.security_context,
-            state_thresholds=self.spec.state_thresholds,
-            serving_spec=self._get_serving_spec(),
-        )
-        job = KubejobRuntime(
-            spec=spec,
-            metadata=self.metadata,
-        )
-        return job

mlrun/runtimes/pod.py

@@ -103,7 +103,6 @@ class KubeResourceSpec(FunctionSpec):
         "preemption_mode",
         "security_context",
         "state_thresholds",
-        "serving_spec",
     ]
     _default_fields_to_strip = FunctionSpec._default_fields_to_strip + [
         "volumes",
@@ -179,7 +178,6 @@ class KubeResourceSpec(FunctionSpec):
         preemption_mode=None,
         security_context=None,
         state_thresholds=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -225,7 +223,6 @@ class KubeResourceSpec(FunctionSpec):
             state_thresholds
             or mlrun.mlconf.function.spec.state_thresholds.default.to_dict()
         )
-        self.serving_spec = serving_spec
         # Termination grace period is internal for runtimes that have a pod termination hook hence it is not in the
         # _dict_fields and doesn't have a setter.
         self._termination_grace_period_seconds = None

mlrun/runtimes/remotesparkjob.py

@@ -58,7 +58,6 @@ class RemoteSparkSpec(KubeResourceSpec):
         preemption_mode=None,
         security_context=None,
         state_thresholds=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -88,7 +87,6 @@ class RemoteSparkSpec(KubeResourceSpec):
             preemption_mode=preemption_mode,
             security_context=security_context,
             state_thresholds=state_thresholds,
-            serving_spec=serving_spec,
         )
         self.provider = provider
 

mlrun/runtimes/sparkjob/spark3job.py

@@ -168,7 +168,6 @@ class Spark3JobSpec(KubeResourceSpec):
         executor_cores=None,
         security_context=None,
         state_thresholds=None,
-        serving_spec=None,
     ):
         super().__init__(
             command=command,
@@ -198,7 +197,6 @@ class Spark3JobSpec(KubeResourceSpec):
             preemption_mode=preemption_mode,
             security_context=security_context,
             state_thresholds=state_thresholds,
-            serving_spec=serving_spec,
         )
 
         self.driver_resources = driver_resources or {}

mlrun/serving/routers.py

@@ -80,10 +80,16 @@ class BaseModelRouter(RouterToDict):
         self._input_path = input_path
         self._result_path = result_path
         self._background_task_check_timestamp = None
-        self._background_task_terminate = False
         self._background_task_current_state = None
         self.kwargs = kwargs
 
+    @property
+    def background_task_reached_terminal_state(self):
+        return (
+            self._background_task_current_state
+            and self._background_task_current_state != "running"
+        )
+
     def parse_event(self, event):
         parsed_event = {}
         try:
@@ -185,35 +191,33 @@ class BaseModelRouter(RouterToDict):
                     background_task.status.state
                     in mlrun.common.schemas.BackgroundTaskState.terminal_states()
                 ):
-                    logger.debug(
+                    logger.info(
                         f"Model endpoint creation task completed with state {background_task.status.state}"
                     )
-                    self._background_task_terminate = True
                 else:  # in progress
-                    logger.debug(
+                    logger.info(
                         f"Model endpoint creation task is still in progress with the current state: "
-                        f"{background_task.status.state}. Events will not be monitored for the next 15 seconds",
+                        f"{background_task.status.state}. Events will not be monitored for the next "
+                        f"{mlrun.mlconf.model_endpoint_monitoring.model_endpoint_creation_check_period} seconds",
                         name=self.name,
                         background_task_check_timestamp=self._background_task_check_timestamp.isoformat(),
                     )
                 return background_task.status.state
             else:
-                logger.debug(
-                    "Model endpoint creation task name not provided",
+                logger.error(
+                    "Model endpoint creation task name not provided. This function is not being monitored.",
                 )
         elif self.context.monitoring_mock:
-            self._background_task_terminate = (
-                True  # If mock monitoring we return success and terminate task check.
-            )
             return mlrun.common.schemas.BackgroundTaskState.succeeded
-        self._background_task_terminate = True  # If mock without monitoring we return failed and terminate task check.
         return mlrun.common.schemas.BackgroundTaskState.failed
 
     def _update_background_task_state(self, event):
-        if not self._background_task_terminate and (
+        if not self.background_task_reached_terminal_state and (
             self._background_task_check_timestamp is None
             or now_date() - self._background_task_check_timestamp
-            >= timedelta(seconds=15)
+            >= timedelta(
+                seconds=mlrun.mlconf.model_endpoint_monitoring.model_endpoint_creation_check_period
+            )
         ):
             self._background_task_current_state = self._get_background_task_status()
         if event.body:

mlrun/serving/server.py

@@ -21,9 +21,8 @@ import os
 import socket
 import traceback
 import uuid
-from typing import Any, Optional, Union
+from typing import Optional, Union
 
-import storey
 from nuclio import Context as NuclioContext
 from nuclio.request import Logger as NuclioLogger
 
@@ -39,10 +38,9 @@ from mlrun.secrets import SecretsStore
 
 from ..common.helpers import parse_versioned_object_uri
 from ..common.schemas.model_monitoring.constants import FileTargetKind
-from ..datastore import DataItem, get_stream_pusher
+from ..datastore import get_stream_pusher
 from ..datastore.store_resources import ResourceCache
 from ..errors import MLRunInvalidArgumentError
-from ..execution import MLClientCtx
 from ..model import ModelObj
 from ..utils import get_caller_globals
 from .states import (
@@ -324,11 +322,7 @@ class GraphServer(ModelObj):
 
     def _process_response(self, context, response, get_body):
         body = response.body
-        if (
-            isinstance(context, MLClientCtx)
-            or isinstance(body, context.Response)
-            or get_body
-        ):
+        if isinstance(body, context.Response) or get_body:
             return body
 
         if body and not isinstance(body, (str, bytes)):
@@ -541,94 +535,6 @@ def v2_serving_init(context, namespace=None):
     _set_callbacks(server, context)
 
 
-async def async_execute_graph(
-    context: MLClientCtx,
-    data: DataItem,
-    batching: bool,
-    batch_size: Optional[int],
-) -> list[Any]:
-    spec = mlrun.utils.get_serving_spec()
-
-    source_filename = spec.get("filename", None)
-    namespace = {}
-    if source_filename:
-        with open(source_filename) as f:
-            exec(f.read(), namespace)
-
-    server = GraphServer.from_dict(spec)
-
-    if config.log_level.lower() == "debug":
-        server.verbose = True
-    context.logger.info_with("Initializing states", namespace=namespace)
-    kwargs = {}
-    if hasattr(context, "is_mock"):
-        kwargs["is_mock"] = context.is_mock
-    server.init_states(
-        context=None,  # this context is expected to be a nuclio context, which we don't have in this flow
-        namespace=namespace,
-        **kwargs,
-    )
-    context.logger.info("Initializing graph steps")
-    server.init_object(namespace)
-
-    context.logger.info_with("Graph was initialized", verbose=server.verbose)
-
-    if server.verbose:
-        context.logger.info(server.to_yaml())
-
-    df = data.as_df()
-
-    responses = []
-
-    async def run(body):
-        event = storey.Event(id=index, body=body)
-        response = await server.run(event, context)
-        responses.append(response)
-
-    if batching and not batch_size:
-        batch_size = len(df)
-
-    batch = []
-    for index, row in df.iterrows():
-        data = row.to_dict()
-        if batching:
-            batch.append(data)
-            if len(batch) == batch_size:
-                await run(batch)
-                batch = []
-        else:
-            await run(data)
-
-    if batch:
-        await run(batch)
-
-    termination_result = server.wait_for_completion()
-    if asyncio.iscoroutine(termination_result):
-        await termination_result
-
-    return responses
-
-
-def execute_graph(
-    context: MLClientCtx,
-    data: DataItem,
-    batching: bool = False,
-    batch_size: Optional[int] = None,
-) -> (list[Any], Any):
-    """
-    Execute graph as a job, from start to finish.
-
-    :param context: The job's execution client context.
-    :param data: The input data to the job, to be pushed into the graph row by row, or in batches.
-    :param batching: Whether to push one or more batches into the graph rather than row by row.
-    :param batch_size: The number of rows to push per batch. If not set, and batching=True, the entire dataset will
-        be pushed into the graph in one batch.
-
-    :return: A list of responses.
-    """
-    return asyncio.run(async_execute_graph(context, data, batching, batch_size))
-
-
 def _set_callbacks(server, context):
     if not server.graph.supports_termination() or not hasattr(context, "platform"):
         return

mlrun/serving/system_steps.py

@@ -314,7 +314,8 @@ class BackgroundTaskStatus(storey.MapClass):
         else:  # in progress
             logger.info(
                 f"Model endpoint creation task is still in progress with the current state: "
-                f"{background_task_state}. Events will not be monitored for the next 15 seconds",
+                f"{background_task_state}. Events will not be monitored for the next "
+                f"{mlrun.mlconf.model_endpoint_monitoring.model_endpoint_creation_check_period} seconds",
                 name=self.name,
                 background_task_check_timestamp=self._background_task_check_timestamp.isoformat(),
             )

mlrun/serving/v2_serving.py

@@ -508,8 +508,8 @@ class V2ModelServer(StepToDict):
                         name=self.name,
                     )
             else:
-                logger.debug(
-                    "Model endpoint creation task name not provided",
+                logger.error(
+                    "Model endpoint creation task name not provided. This function is not being monitored.",
                 )
 
 

mlrun/utils/helpers.py

@@ -911,7 +911,7 @@ def enrich_image_url(
     if is_mlrun_image and "mlrun/ml-base" in image_url:
         if tag:
             if mlrun.utils.helpers.validate_component_version_compatibility(
-                "mlrun-client", "1.10.0", mlrun_client_version=tag
+                "mlrun-client", "1.10.0-rc0", mlrun_client_version=tag
             ):
                 warnings.warn(
                     "'mlrun/ml-base' image is deprecated in 1.10.0 and will be removed in 1.12.0, "

mlrun/utils/version/version.json

@@ -1,4 +1,4 @@
 {
-  "git_commit": "fede26558b2c8db736315ad1f48e15e3ce2f387d",
-  "version": "1.10.0-rc8"
+  "git_commit": "7711525d2af418d6991128a0d3253094584ddedc",
+  "version": "1.10.0-rc10"
 }

{mlrun-1.10.0rc8.dist-info → mlrun-1.10.0rc10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlrun
-Version: 1.10.0rc8
+Version: 1.10.0rc10
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
@@ -9,7 +9,6 @@ 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
 Classifier: Operating System :: POSIX :: Linux
 Classifier: Operating System :: Microsoft :: Windows
 Classifier: Operating System :: MacOS
@@ -85,11 +84,11 @@ Requires-Dist: avro~=1.11; extra == "kafka"
 Provides-Extra: redis
 Requires-Dist: redis~=4.3; extra == "redis"
 Provides-Extra: mlflow
-Requires-Dist: mlflow~=2.16; extra == "mlflow"
+Requires-Dist: mlflow~=2.22; extra == "mlflow"
 Provides-Extra: databricks-sdk
 Requires-Dist: databricks-sdk~=0.20.0; extra == "databricks-sdk"
 Provides-Extra: sqlalchemy
-Requires-Dist: sqlalchemy~=1.4; extra == "sqlalchemy"
+Requires-Dist: sqlalchemy~=2.0; extra == "sqlalchemy"
 Provides-Extra: dask
 Requires-Dist: dask~=2024.12.1; python_version >= "3.11" and extra == "dask"
 Requires-Dist: distributed~=2024.12.1; python_version >= "3.11" and extra == "dask"
@@ -112,8 +111,10 @@ Requires-Dist: objgraph~=3.6; extra == "api"
 Requires-Dist: igz-mgmt~=0.4.1; extra == "api"
 Requires-Dist: humanfriendly~=10.0; extra == "api"
 Requires-Dist: fastapi~=0.115.6; extra == "api"
-Requires-Dist: sqlalchemy~=1.4; extra == "api"
+Requires-Dist: sqlalchemy~=2.0; extra == "api"
+Requires-Dist: sqlalchemy-utils~=0.41.2; extra == "api"
 Requires-Dist: pymysql~=1.1; extra == "api"
+Requires-Dist: psycopg2-binary~=2.9; extra == "api"
 Requires-Dist: alembic~=1.14; extra == "api"
 Requires-Dist: timelength~=1.1; extra == "api"
 Requires-Dist: memray~=1.12; sys_platform != "win32" and extra == "api"
@@ -141,7 +142,7 @@ Requires-Dist: google-cloud-storage==2.14.0; extra == "all"
 Requires-Dist: google-cloud==0.34; extra == "all"
 Requires-Dist: graphviz~=0.20.0; extra == "all"
 Requires-Dist: kafka-python~=2.1.0; extra == "all"
-Requires-Dist: mlflow~=2.16; extra == "all"
+Requires-Dist: mlflow~=2.22; extra == "all"
 Requires-Dist: msrest~=0.6.21; extra == "all"
 Requires-Dist: oss2==2.18.1; extra == "all"
 Requires-Dist: ossfs==2023.12.0; extra == "all"
@@ -150,7 +151,7 @@ Requires-Dist: pyopenssl>=23; extra == "all"
 Requires-Dist: redis~=4.3; extra == "all"
 Requires-Dist: s3fs<2024.7,>=2023.9.2; extra == "all"
 Requires-Dist: snowflake-connector-python~=3.7; extra == "all"
-Requires-Dist: sqlalchemy~=1.4; extra == "all"
+Requires-Dist: sqlalchemy~=2.0; extra == "all"
 Requires-Dist: taos-ws-py==0.3.2; extra == "all"
 Provides-Extra: complete
 Requires-Dist: adlfs==2023.9.0; extra == "complete"
@@ -172,7 +173,7 @@ Requires-Dist: google-cloud-storage==2.14.0; extra == "complete"
 Requires-Dist: google-cloud==0.34; extra == "complete"
 Requires-Dist: graphviz~=0.20.0; extra == "complete"
 Requires-Dist: kafka-python~=2.1.0; extra == "complete"
-Requires-Dist: mlflow~=2.16; extra == "complete"
+Requires-Dist: mlflow~=2.22; extra == "complete"
 Requires-Dist: msrest~=0.6.21; extra == "complete"
 Requires-Dist: oss2==2.18.1; extra == "complete"
 Requires-Dist: ossfs==2023.12.0; extra == "complete"
@@ -181,7 +182,7 @@ Requires-Dist: pyopenssl>=23; extra == "complete"
 Requires-Dist: redis~=4.3; extra == "complete"
 Requires-Dist: s3fs<2024.7,>=2023.9.2; extra == "complete"
 Requires-Dist: snowflake-connector-python~=3.7; extra == "complete"
-Requires-Dist: sqlalchemy~=1.4; extra == "complete"
+Requires-Dist: sqlalchemy~=2.0; extra == "complete"
 Requires-Dist: taos-ws-py==0.3.2; extra == "complete"
 Provides-Extra: complete-api
 Requires-Dist: adlfs==2023.9.0; extra == "complete-api"
@@ -212,20 +213,22 @@ Requires-Dist: humanfriendly~=10.0; extra == "complete-api"
 Requires-Dist: igz-mgmt~=0.4.1; extra == "complete-api"
 Requires-Dist: kafka-python~=2.1.0; extra == "complete-api"
 Requires-Dist: memray~=1.12; sys_platform != "win32" and extra == "complete-api"
-Requires-Dist: mlflow~=2.16; extra == "complete-api"
+Requires-Dist: mlflow~=2.22; extra == "complete-api"
 Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.4; extra == "complete-api"
 Requires-Dist: msrest~=0.6.21; extra == "complete-api"
 Requires-Dist: objgraph~=3.6; extra == "complete-api"
 Requires-Dist: oss2==2.18.1; extra == "complete-api"
 Requires-Dist: ossfs==2023.12.0; extra == "complete-api"
 Requires-Dist: plotly~=5.23; extra == "complete-api"
+Requires-Dist: psycopg2-binary~=2.9; extra == "complete-api"
 Requires-Dist: pydantic<2,>=1; extra == "complete-api"
 Requires-Dist: pymysql~=1.1; extra == "complete-api"
 Requires-Dist: pyopenssl>=23; extra == "complete-api"
 Requires-Dist: redis~=4.3; extra == "complete-api"
 Requires-Dist: s3fs<2024.7,>=2023.9.2; extra == "complete-api"
 Requires-Dist: snowflake-connector-python~=3.7; extra == "complete-api"
-Requires-Dist: sqlalchemy~=1.4; extra == "complete-api"
+Requires-Dist: sqlalchemy-utils~=0.41.2; extra == "complete-api"
+Requires-Dist: sqlalchemy~=2.0; extra == "complete-api"
 Requires-Dist: taos-ws-py==0.3.2; extra == "complete-api"
 Requires-Dist: timelength~=1.1; extra == "complete-api"
 Requires-Dist: uvicorn~=0.32.1; extra == "complete-api"
@@ -249,7 +252,7 @@ Dynamic: summary
 [![PyPI version fury.io](https://badge.fury.io/py/mlrun.svg)](https://pypi.python.org/pypi/mlrun/)
 [![Documentation](https://readthedocs.org/projects/mlrun/badge/?version=latest)](https://mlrun.readthedocs.io/en/latest/?badge=latest)
 [![Ruff](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json)](https://github.com/astral-sh/ruff)
-![GitHub commit activity](https://img.shields.io/github/commit-activity/w/mlrun/mlrun)
+[![GitHub commit activity](https://img.shields.io/github/commit-activity/w/mlrun/mlrun)](https://github.com/mlrun/mlrun/commits/main)
 [![GitHub release (latest SemVer)](https://img.shields.io/github/v/release/mlrun/mlrun?sort=semver)](https://github.com/mlrun/mlrun/releases)
 [![Join MLOps Live](https://img.shields.io/badge/slack-join_chat-white.svg?logo=slack&style=social)](https://mlopslive.slack.com)