mlrun 1.10.0rc29__tar.gz → 1.10.0rc31__tar.gz

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/Makefile

@@ -1036,9 +1036,7 @@ upgrade-mlrun-test-deps-lock: ## Upgrade mlrun test locked requirements file
 		dockerfiles/test/requirements.txt \
 		dev-requirements.txt \
 		$(MLRUN_UV_UPGRADE_FLAG) \
-		--python-version $(MLRUN_PYTHON_VERSION) \
-		--constraint dockerfiles/constraints-py$(MLRUN_PYTHON_VERSION).txt \
-		--output-file dockerfiles/test/locked-requirements_${MLRUN_PYTHON_VERSION}.txt
+		--output-file dockerfiles/test/locked-requirements.txt
 
 .PHONY: upgrade-mlrun-system-test-deps-lock
 upgrade-mlrun-system-test-deps-lock: ## Upgrade mlrun system test locked requirements file
@@ -1052,15 +1050,6 @@ upgrade-mlrun-system-test-deps-lock: ## Upgrade mlrun system test locked require
 		--output-file dockerfiles/test-system/locked-requirements.txt
 
 
-upgrade-mlrun-test-deps-lock-all: upgrade-mlrun-test-deps-lock-3.11 upgrade-mlrun-test-deps-lock-3.9
-
-upgrade-mlrun-test-deps-lock-3.11:
-	$(MAKE) upgrade-mlrun-test-deps-lock MLRUN_PYTHON_VERSION=3.11
-
-upgrade-mlrun-test-deps-lock-3.9:
-	$(MAKE) upgrade-mlrun-test-deps-lock MLRUN_PYTHON_VERSION=3.9
-
-
 upgrade-mlrun-kfp-deps-lock: ## Upgrade mlrun-kfp locked requirements file
 	uv pip compile \
 		requirements.txt \
@@ -1077,7 +1066,7 @@ upgrade-mlrun-deps-lock: ## Upgrade mlrun-* locked requirements file
 		upgrade-mlrun-jupyter-deps-lock \
 		upgrade-mlrun-gpu-deps-lock \
 		upgrade-mlrun-kfp-deps-lock \
-		upgrade-mlrun-test-deps-lock-all \
+		upgrade-mlrun-test-deps-lock \
 		upgrade-mlrun-system-test-deps-lock
 
 

{mlrun-1.10.0rc29/mlrun.egg-info → mlrun-1.10.0rc31}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlrun
-Version: 1.10.0rc29
+Version: 1.10.0rc31
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
@@ -22,7 +22,8 @@ Requires-Python: >=3.9, <3.12
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: urllib3>=1.26.20
-Requires-Dist: v3io-frames>=0.10.16
+Requires-Dist: v3io-frames~=0.10.16; python_version < "3.11"
+Requires-Dist: v3io-frames~=0.13.11; python_version >= "3.11"
 Requires-Dist: GitPython>=3.1.41,~=3.1
 Requires-Dist: aiohttp~=3.11
 Requires-Dist: aiohttp-retry~=2.9
@@ -43,15 +44,15 @@ Requires-Dist: semver~=3.0
 Requires-Dist: dependency-injector~=4.41
 Requires-Dist: fsspec<=2025.7.0,>=2025.5.1
 Requires-Dist: v3iofs~=0.1.17
-Requires-Dist: storey~=1.10.13
+Requires-Dist: storey~=1.10.14
 Requires-Dist: inflection~=0.5.0
 Requires-Dist: python-dotenv~=1.0
 Requires-Dist: setuptools>=75.2
 Requires-Dist: deprecated~=1.2
 Requires-Dist: jinja2>=3.1.6,~=3.1
 Requires-Dist: orjson<4,>=3.9.15
-Requires-Dist: mlrun-pipelines-kfp-common~=0.5.8
-Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.7
+Requires-Dist: mlrun-pipelines-kfp-common~=0.5.9
+Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.8
 Requires-Dist: docstring_parser~=0.16
 Requires-Dist: aiosmtplib~=3.0
 Requires-Dist: deepdiff<9.0.0,>=8.6.1
@@ -122,7 +123,7 @@ Requires-Dist: timelength~=1.1; extra == "api"
 Requires-Dist: memray~=1.12; sys_platform != "win32" and extra == "api"
 Requires-Dist: aiosmtplib~=3.0; extra == "api"
 Requires-Dist: pydantic<2,>=1; extra == "api"
-Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.7; extra == "api"
+Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.8; extra == "api"
 Provides-Extra: all
 Requires-Dist: adlfs==2024.12.0; extra == "all"
 Requires-Dist: aiobotocore<2.16,>=2.5.0; extra == "all"
@@ -214,7 +215,7 @@ 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~=2.22; extra == "complete-api"
-Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.7; extra == "complete-api"
+Requires-Dist: mlrun-pipelines-kfp-v1-8~=0.5.8; extra == "complete-api"
 Requires-Dist: msrest~=0.6.21; extra == "complete-api"
 Requires-Dist: objgraph~=3.6; extra == "complete-api"
 Requires-Dist: oss2==2.18.4; extra == "complete-api"

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/dockerfiles/mlrun-api/requirements.txt

@@ -17,4 +17,4 @@ memray~=1.12; sys_platform != 'win32'
 aiosmtplib~=3.0
 # until https://github.com/fastapi/fastapi/issues/10360 is solved
 pydantic>=1,<2
-mlrun-pipelines-kfp-v1-8~=0.5.7
+mlrun-pipelines-kfp-v1-8~=0.5.8

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/artifacts/llm_prompt.py

@@ -41,7 +41,7 @@ class LLMPromptArtifactSpec(ArtifactSpec):
         prompt_template: Optional[list[dict]] = None,
         prompt_path: Optional[str] = None,
         prompt_legend: Optional[dict] = None,
-        model_configuration: Optional[dict] = None,
+        invocation_config: Optional[dict] = None,
         description: Optional[str] = None,
         target_path: Optional[str] = None,
         **kwargs,
@@ -68,13 +68,11 @@ class LLMPromptArtifactSpec(ArtifactSpec):
 
         self.prompt_template = prompt_template
         self.prompt_legend = prompt_legend
-        if model_configuration is not None and not isinstance(
-            model_configuration, dict
-        ):
+        if invocation_config is not None and not isinstance(invocation_config, dict):
             raise mlrun.errors.MLRunInvalidArgumentError(
                 "LLMPromptArtifact model_configuration must be a dictionary or None"
             )
-        self.model_configuration = model_configuration or {}
+        self.model_configuration = invocation_config or {}
         self.description = description
         self._model_artifact = (
             model_artifact
@@ -177,7 +175,7 @@ class LLMPromptArtifact(Artifact):
         prompt_template: Optional[list[dict]] = None,
         prompt_path: Optional[str] = None,
         prompt_legend: Optional[dict] = None,
-        model_configuration: Optional[dict] = None,
+        invocation_config: Optional[dict] = None,
         description: Optional[str] = None,
         target_path=None,
         **kwargs,
@@ -187,7 +185,7 @@ class LLMPromptArtifact(Artifact):
             prompt_path=prompt_path,
             prompt_legend=prompt_legend,
             model_artifact=model_artifact,
-            model_configuration=model_configuration,
+            invocation_config=invocation_config,
             target_path=target_path,
             description=description,
         )

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/common/schemas/model_monitoring/constants.py

@@ -196,6 +196,10 @@ class WriterEventKind(MonitoringStrEnum):
     RESULT = "result"
     STATS = "stats"
 
+    @classmethod
+    def user_app_outputs(cls):
+        return [cls.METRIC, cls.RESULT]
+
 
 class ControllerEvent(MonitoringStrEnum):
     KIND = "kind"
@@ -304,6 +308,7 @@ class FileTargetKind:
     MONITORING_APPLICATION = "monitoring_application"
     ERRORS = "errors"
     STATS = "stats"
+    PARQUET_STATS = "parquet_stats"
     LAST_REQUEST = "last_request"
 
 

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/config.py

@@ -649,6 +649,13 @@ default_config = {
                 "max_replicas": 1,
             },
         },
+        "writer_graph": {
+            "max_events": 1000,
+            "flush_after_seconds": 30,
+            "writer_version": "v1",  # v1 is the sync version while v2 is async
+            "parquet_batching_max_events": 10,
+            "parquet_batching_timeout_secs": 30,
+        },
         # Store prefixes are used to handle model monitoring storing policies based on project and kind, such as events,
         # stream, and endpoints.
         "store_prefixes": {

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/execution.py

@@ -917,7 +917,7 @@ class MLClientCtx:
         prompt_path: Optional[str] = None,
         prompt_legend: Optional[dict] = None,
         model_artifact: Union[ModelArtifact, str] = None,
-        model_configuration: Optional[dict] = None,
+        invocation_config: Optional[dict] = None,
         description: Optional[str] = None,
         target_path: Optional[str] = None,
         artifact_path: Optional[str] = None,
@@ -997,7 +997,7 @@ class MLClientCtx:
                with the place-holder name. "description" will point to explanation of what that placeholder represents.
                Useful for documenting and clarifying dynamic parts of the prompt.
         :param model_artifact: Reference to the parent model (either `ModelArtifact` or model URI string).
-        :param model_configuration: Dictionary of generation parameters (e.g., temperature, max_tokens).
+        :param invocation_config: Dictionary of generation parameters (e.g., temperature, max_tokens).
         :param description:   Optional description of the prompt.
         :param target_path:   Absolute target path (instead of using artifact_path + local_path)
         :param artifact_path: Target artifact path (when not using the default)
@@ -1023,7 +1023,7 @@ class MLClientCtx:
             prompt_path=prompt_path,
             prompt_legend=prompt_legend,
             model_artifact=model_artifact,
-            model_configuration=model_configuration,
+            invocation_config=invocation_config,
             target_path=target_path,
             description=description,
             **kwargs,

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/model_monitoring/db/_stats.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 import abc
 import json
+import typing
 from abc import abstractmethod
 from datetime import datetime, timezone
 from typing import cast
@@ -73,7 +74,7 @@ class ModelMonitoringStatsFile(abc.ABC):
                 path=self._item.url,
             )
 
-    def read(self) -> tuple[dict, datetime]:
+    def read(self) -> tuple[dict, typing.Optional[datetime]]:
         """
         Read the stats data and timestamp saved in file
         :return: tuple[dict, str] dictionary with stats data and timestamp saved in file
@@ -99,13 +100,13 @@ class ModelMonitoringStatsFile(abc.ABC):
             ):
                 raise
 
-            logger.exception(
+            logger.warning(
                 "The Stats file was not found. It should have been created "
                 "as a part of the model endpoint's creation",
                 path=self._path,
                 error=err,
             )
-            raise
+            return {}, None
 
     def write(self, stats: dict, timestamp: datetime) -> None:
         """

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/model_monitoring/db/tsdb/base.py

@@ -60,6 +60,16 @@ class TSDBConnector(ABC):
         """
         pass
 
+    def apply_writer_steps(self, graph, after, **kwargs) -> None:
+        """
+        Apply TSDB steps on the provided writer graph. Throughout these steps, the graph stores metrics / results.
+        This data is being used by mlrun UI and the monitoring dashboards in grafana.
+        There are 2 different key metric dictionaries that are being generated throughout these steps:
+        - metrics (user-defined metrics) - model monitoring application metrics
+        - results (user-defined results) - model monitoring application results
+        """
+        pass
+
     @abstractmethod
     def handle_model_error(self, graph, **kwargs) -> None:
         """
@@ -783,3 +793,6 @@ class TSDBConnector(ABC):
             )
         )
         return mm_schemas.ModelEndpointDriftValues(values=values)
+
+    def add_pre_writer_steps(self, graph, after):
+        return None

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/model_monitoring/db/tsdb/tdengine/tdengine_connector.py

@@ -22,6 +22,7 @@ import taosws
 import mlrun.common.schemas.model_monitoring as mm_schemas
 import mlrun.common.types
 import mlrun.model_monitoring.db.tsdb.tdengine.schemas as tdengine_schemas
+from mlrun.config import config
 from mlrun.datastore.datastore_profile import DatastoreProfile
 from mlrun.model_monitoring.db import TSDBConnector
 from mlrun.model_monitoring.db.tsdb.tdengine.tdengine_connection import (
@@ -277,6 +278,65 @@ class TDEngineConnector(TSDBConnector):
             after="ProcessBeforeTDEngine",
         )
 
+    def add_pre_writer_steps(self, graph, after):
+        return graph.add_step(
+            "mlrun.model_monitoring.db.tsdb.tdengine.writer_graph_steps.ProcessBeforeTDEngine",
+            name="ProcessBeforeTDEngine",
+            after=after,
+        )
+
+    def apply_writer_steps(self, graph, after, **kwargs) -> None:
+        graph.add_step(
+            "mlrun.datastore.storeytargets.TDEngineStoreyTarget",
+            name="tsdb_metrics",
+            after=after,
+            url=f"ds://{self._tdengine_connection_profile.name}",
+            supertable=self.tables[mm_schemas.TDEngineSuperTables.METRICS].super_table,
+            table_col=mm_schemas.EventFieldType.TABLE_COLUMN,
+            time_col=mm_schemas.WriterEvent.END_INFER_TIME,
+            database=self.database,
+            graph_shape="cylinder",
+            columns=[
+                mm_schemas.WriterEvent.START_INFER_TIME,
+                mm_schemas.MetricData.METRIC_VALUE,
+            ],
+            tag_cols=[
+                mm_schemas.WriterEvent.ENDPOINT_ID,
+                mm_schemas.WriterEvent.APPLICATION_NAME,
+                mm_schemas.MetricData.METRIC_NAME,
+            ],
+            max_events=config.model_endpoint_monitoring.writer_graph.max_events,
+            flush_after_seconds=config.model_endpoint_monitoring.writer_graph.flush_after_seconds,
+        )
+
+        graph.add_step(
+            "mlrun.datastore.storeytargets.TDEngineStoreyTarget",
+            name="tsdb_app_results",
+            after=after,
+            url=f"ds://{self._tdengine_connection_profile.name}",
+            supertable=self.tables[
+                mm_schemas.TDEngineSuperTables.APP_RESULTS
+            ].super_table,
+            table_col=mm_schemas.EventFieldType.TABLE_COLUMN,
+            time_col=mm_schemas.WriterEvent.END_INFER_TIME,
+            database=self.database,
+            graph_shape="cylinder",
+            columns=[
+                mm_schemas.WriterEvent.START_INFER_TIME,
+                mm_schemas.ResultData.RESULT_VALUE,
+                mm_schemas.ResultData.RESULT_STATUS,
+                mm_schemas.ResultData.RESULT_EXTRA_DATA,
+            ],
+            tag_cols=[
+                mm_schemas.WriterEvent.ENDPOINT_ID,
+                mm_schemas.WriterEvent.APPLICATION_NAME,
+                mm_schemas.ResultData.RESULT_NAME,
+                mm_schemas.ResultData.RESULT_KIND,
+            ],
+            max_events=config.model_endpoint_monitoring.writer_graph.max_events,
+            flush_after_seconds=config.model_endpoint_monitoring.writer_graph.flush_after_seconds,
+        )
+
     def handle_model_error(
         self,
         graph,

mlrun-1.10.0rc31/mlrun/model_monitoring/db/tsdb/tdengine/writer_graph_steps.py

@@ -0,0 +1,51 @@
+# Copyright 2025 Iguazio
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from datetime import datetime
+
+import mlrun.common.schemas.model_monitoring as mm_schemas
+import mlrun.feature_store.steps
+from mlrun.utils import logger
+
+
+class ProcessBeforeTDEngine(mlrun.feature_store.steps.MapClass):
+    def __init__(self, **kwargs):
+        """
+        Process the data before writing to TDEngine. This step create the table name.
+
+        :returns: Event as a dictionary which will be written into the TDEngine Metrics/Results tables.
+        """
+        super().__init__(**kwargs)
+
+    def do(self, event):
+        logger.info("Process event before writing to TDEngine", event=event)
+        kind = event.get("kind")
+        table_name = (
+            f"{event[mm_schemas.WriterEvent.ENDPOINT_ID]}_"
+            f"{event[mm_schemas.WriterEvent.APPLICATION_NAME]}"
+        )
+        if kind == mm_schemas.WriterEventKind.RESULT:
+            # Write a new result
+            event[mm_schemas.EventFieldType.TABLE_COLUMN] = (
+                f"{table_name}_{event[mm_schemas.ResultData.RESULT_NAME]}"
+            ).replace("-", "_")
+        elif kind == mm_schemas.WriterEventKind.METRIC:
+            # Write a new metric
+            event[mm_schemas.EventFieldType.TABLE_COLUMN] = (
+                f"{table_name}_{event[mm_schemas.MetricData.METRIC_NAME]}"
+            ).replace("-", "_")
+        event[mm_schemas.WriterEvent.START_INFER_TIME] = datetime.fromisoformat(
+            event[mm_schemas.WriterEvent.START_INFER_TIME]
+        )
+        return event

{mlrun-1.10.0rc29 → mlrun-1.10.0rc31}/mlrun/model_monitoring/db/tsdb/v3io/v3io_connector.py

@@ -25,6 +25,7 @@ import mlrun.common.schemas.model_monitoring as mm_schemas
 import mlrun.feature_store.steps
 import mlrun.utils.v3io_clients
 from mlrun.common.schemas import EventFieldType
+from mlrun.config import config
 from mlrun.model_monitoring.db import TSDBConnector
 from mlrun.model_monitoring.helpers import get_invocations_fqn, get_start_end
 from mlrun.utils import logger
@@ -369,6 +370,49 @@ class V3IOTSDBConnector(TSDBConnector):
         apply_storey_filter()
         apply_tsdb_target(name="tsdb3", after="FilterNotNone")
 
+    def apply_writer_steps(self, graph, after, **kwargs) -> None:
+        graph.add_step(
+            "storey.TSDBTarget",
+            name="tsdb_metrics",
+            after=after,
+            path=f"{self.container}/{self.tables[mm_schemas.V3IOTSDBTables.METRICS]}",
+            time_col=mm_schemas.WriterEvent.END_INFER_TIME,
+            container=self.container,
+            v3io_frames=self.v3io_framesd,
+            infer_columns_from_data=True,
+            graph_shape="cylinder",
+            index_cols=[
+                mm_schemas.WriterEvent.APPLICATION_NAME,
+                mm_schemas.WriterEvent.ENDPOINT_NAME,
+                mm_schemas.WriterEvent.ENDPOINT_ID,
+                mm_schemas.MetricData.METRIC_NAME,
+            ],
+            max_events=config.model_endpoint_monitoring.writer_graph.max_events,
+            flush_after_seconds=config.model_endpoint_monitoring.writer_graph.flush_after_seconds,
+            key=mm_schemas.EventFieldType.ENDPOINT_ID,
+        )
+
+        graph.add_step(
+            "storey.TSDBTarget",
+            name="tsdb_app_results",
+            after=after,
+            path=f"{self.container}/{self.tables[mm_schemas.V3IOTSDBTables.APP_RESULTS]}",
+            time_col=mm_schemas.WriterEvent.END_INFER_TIME,
+            container=self.container,
+            v3io_frames=self.v3io_framesd,
+            infer_columns_from_data=True,
+            graph_shape="cylinder",
+            index_cols=[
+                mm_schemas.WriterEvent.APPLICATION_NAME,
+                mm_schemas.WriterEvent.ENDPOINT_NAME,
+                mm_schemas.WriterEvent.ENDPOINT_ID,
+                mm_schemas.ResultData.RESULT_NAME,
+            ],
+            max_events=config.model_endpoint_monitoring.writer_graph.max_events,
+            flush_after_seconds=config.model_endpoint_monitoring.writer_graph.flush_after_seconds,
+            key=mm_schemas.EventFieldType.ENDPOINT_ID,
+        )
+
     def handle_model_error(
         self,
         graph,