mlrun 1.8.0rc30__py3-none-any.whl → 1.8.0rc31__py3-none-any.whl

mlrun/__init__.py

@@ -26,7 +26,6 @@ __all__ = [
     "VolumeMount",
 ]
 
-import collections
 from os import environ, path
 from typing import Optional
 
@@ -215,40 +214,8 @@ def set_env_from_file(env_file: str, return_dict: bool = False) -> Optional[dict
     if None in env_vars.values():
         raise MLRunInvalidArgumentError("env file lines must be in the form key=value")
 
-    ordered_env_vars = order_env_vars(env_vars)
-    for key, value in ordered_env_vars.items():
+    for key, value in env_vars.items():
         environ[key] = value
 
     mlconf.reload()  # reload mlrun configuration
-    return ordered_env_vars if return_dict else None
-
-
-def order_env_vars(env_vars: dict[str, str]) -> dict[str, str]:
-    """
-    Order and process environment variables by first handling specific ordered keys,
-    then processing the remaining keys in the given dictionary.
-
-    The function ensures that environment variables defined in the `ordered_keys` list
-    are added to the result dictionary first. Any other environment variables from
-    `env_vars` are then added in the order they appear in the input dictionary.
-
-    :param env_vars: A dictionary where each key is the name of an environment variable (str),
-                      and each value is the corresponding environment variable value (str).
-    :return: A dictionary with the processed environment variables, ordered with the specific
-             keys first, followed by the rest in their original order.
-    """
-    ordered_keys = mlconf.get_ordered_keys()
-
-    ordered_env_vars = collections.OrderedDict()
-
-    # First, add the ordered keys to the dictionary
-    for key in ordered_keys:
-        if key in env_vars:
-            ordered_env_vars[key] = env_vars[key]
-
-    # Then, add the remaining keys (those not in ordered_keys)
-    for key, value in env_vars.items():
-        if key not in ordered_keys:
-            ordered_env_vars[key] = value
-
-    return ordered_env_vars
+    return env_vars if return_dict else None

mlrun/api/schemas/__init__.py

@@ -193,9 +193,7 @@ FeatureValues = DeprecationHelper(mlrun.common.schemas.FeatureValues)
 GrafanaColumn = DeprecationHelper(
     mlrun.common.schemas.model_monitoring.grafana.GrafanaColumn
 )
-GrafanaDataPoint = DeprecationHelper(
-    mlrun.common.schemas.model_monitoring.grafana.GrafanaDataPoint
-)
+
 GrafanaNumberColumn = DeprecationHelper(
     mlrun.common.schemas.model_monitoring.grafana.GrafanaNumberColumn
 )
@@ -205,9 +203,6 @@ GrafanaStringColumn = DeprecationHelper(
 GrafanaTable = DeprecationHelper(
     mlrun.common.schemas.model_monitoring.grafana.GrafanaTable
 )
-GrafanaTimeSeriesTarget = DeprecationHelper(
-    mlrun.common.schemas.model_monitoring.grafana.GrafanaTimeSeriesTarget
-)
 ModelEndpoint = DeprecationHelper(mlrun.common.schemas.ModelEndpoint)
 ModelEndpointList = DeprecationHelper(mlrun.common.schemas.ModelEndpointList)
 ModelEndpointMetadata = DeprecationHelper(mlrun.common.schemas.ModelEndpointMetadata)

mlrun/common/runtimes/constants.py

@@ -194,6 +194,10 @@ class RunStates:
             # TODO: add aborting state once we have it
         ]
 
+    @staticmethod
+    def notification_states():
+        return RunStates.terminal_states() + [RunStates.running]
+
     @staticmethod
     def run_state_to_pipeline_run_status(run_state: str):
         if not run_state:

mlrun/common/schemas/__init__.py

@@ -140,11 +140,9 @@ from .model_monitoring import (
     FeatureSetFeatures,
     FeatureValues,
     GrafanaColumn,
-    GrafanaDataPoint,
     GrafanaNumberColumn,
     GrafanaStringColumn,
     GrafanaTable,
-    GrafanaTimeSeriesTarget,
     ModelEndpoint,
     ModelEndpointCreationStrategy,
     ModelEndpointList,

mlrun/common/schemas/model_monitoring/__init__.py

@@ -51,11 +51,9 @@ from .constants import (
 from .grafana import (
     GrafanaColumn,
     GrafanaColumnType,
-    GrafanaDataPoint,
     GrafanaNumberColumn,
     GrafanaStringColumn,
     GrafanaTable,
-    GrafanaTimeSeriesTarget,
 )
 from .model_endpoints import (
     Features,

mlrun/common/schemas/model_monitoring/constants.py

@@ -250,11 +250,6 @@ class TSDBTarget(MonitoringStrEnum):
     TDEngine = "tdengine"
 
 
-class DefaultProfileName(StrEnum):
-    STREAM = "mm-infra-stream"
-    TSDB = "mm-infra-tsdb"
-
-
 class ProjectSecretKeys:
     ACCESS_KEY = "MODEL_MONITORING_ACCESS_KEY"
     TSDB_PROFILE_NAME = "TSDB_PROFILE_NAME"
@@ -473,8 +468,8 @@ FQN_REGEX = re.compile(FQN_PATTERN)
 
 # refer to `mlrun.utils.regex.project_name`
 PROJECT_PATTERN = r"^[a-z0-9]([a-z0-9-]{0,61}[a-z0-9])?$"
-
 MODEL_ENDPOINT_ID_PATTERN = r"^[a-zA-Z0-9_-]+$"
+RESULT_NAME_PATTERN = r"[a-zA-Z_][a-zA-Z0-9_]*"
 
 INTERSECT_DICT_KEYS = {
     ModelEndpointMonitoringMetricType.METRIC: "intersect_metrics",

mlrun/common/schemas/model_monitoring/grafana.py

@@ -46,14 +46,20 @@ class GrafanaTable(BaseModel):
         self.rows.append(list(args))
 
 
-class GrafanaDataPoint(BaseModel):
-    value: float
-    timestamp: int  # Unix timestamp in milliseconds
-
-
-class GrafanaTimeSeriesTarget(BaseModel):
-    target: str
-    datapoints: list[tuple[float, int]] = []
-
-    def add_data_point(self, data_point: GrafanaDataPoint):
-        self.datapoints.append((data_point.value, data_point.timestamp))
+class GrafanaModelEndpointsTable(GrafanaTable):
+    def __init__(self):
+        columns = self._init_columns()
+        super().__init__(columns=columns)
+
+    @staticmethod
+    def _init_columns():
+        return [
+            GrafanaColumn(text="endpoint_id", type=GrafanaColumnType.STRING),
+            GrafanaColumn(text="endpoint_name", type=GrafanaColumnType.STRING),
+            GrafanaColumn(text="endpoint_function", type=GrafanaColumnType.STRING),
+            GrafanaColumn(text="endpoint_model", type=GrafanaColumnType.STRING),
+            GrafanaColumn(text="endpoint_model_class", type=GrafanaColumnType.STRING),
+            GrafanaColumn(text="error_count", type=GrafanaColumnType.NUMBER),
+            GrafanaColumn(text="drift_status", type=GrafanaColumnType.NUMBER),
+            GrafanaColumn(text="sampling_percentage", type=GrafanaColumnType.NUMBER),
+        ]

mlrun/config.py

@@ -1366,35 +1366,6 @@ class Config:
             ver in mlrun.mlconf.ce.mode for ver in ["lite", "full"]
         )
 
-    def get_s3_storage_options(self) -> dict[str, typing.Any]:
-        """
-        Generate storage options dictionary as required for handling S3 path in fsspec. The model monitoring stream
-        graph uses this method for generating the storage options for S3 parquet target path.
-        :return: A storage options dictionary in which each key-value pair  represents a particular configuration,
-        such as endpoint_url or aws access key.
-        """
-        key = mlrun.get_secret_or_env("AWS_ACCESS_KEY_ID")
-        secret = mlrun.get_secret_or_env("AWS_SECRET_ACCESS_KEY")
-
-        force_non_anonymous = mlrun.get_secret_or_env("S3_NON_ANONYMOUS")
-        profile = mlrun.get_secret_or_env("AWS_PROFILE")
-
-        storage_options = dict(
-            anon=not (force_non_anonymous or (key and secret)),
-            key=key,
-            secret=secret,
-        )
-
-        endpoint_url = mlrun.get_secret_or_env("S3_ENDPOINT_URL")
-        if endpoint_url:
-            client_kwargs = {"endpoint_url": endpoint_url}
-            storage_options["client_kwargs"] = client_kwargs
-
-        if profile:
-            storage_options["profile"] = profile
-
-        return storage_options
-
     def is_explicit_ack_enabled(self) -> bool:
         return self.httpdb.nuclio.explicit_ack == "enabled" and (
             not self.nuclio_version
@@ -1402,13 +1373,6 @@ class Config:
             >= semver.VersionInfo.parse("1.12.10")
         )
 
-    @staticmethod
-    def get_ordered_keys():
-        # Define the keys to process first
-        return [
-            "MLRUN_HTTPDB__HTTP__VERIFY"  # Ensure this key is processed first for proper connection setup
-        ]
-
 
 # Global configuration
 config = Config.from_dict(default_config)
@@ -1626,6 +1590,15 @@ def read_env(env=None, prefix=env_prefix):
     # The default function pod resource values are of type str; however, when reading from environment variable numbers,
     # it converts them to type int if contains only number, so we want to convert them to str.
     _convert_resources_to_str(config)
+
+    # If the environment variable MLRUN_HTTPDB__HTTP__VERIFY is set, we ensure SSL verification settings take precedence
+    # by moving the 'httpdb' configuration to the beginning of the config dictionary.
+    # This ensures that SSL verification is applied before other settings.
+    if "MLRUN_HTTPDB__HTTP__VERIFY" in env:
+        httpdb = config.pop("httpdb", None)
+        if httpdb:
+            config = {"httpdb": httpdb, **config}
+
     return config
 
 

mlrun/datastore/storeytargets.py

@@ -42,9 +42,21 @@ def get_url_and_storage_options(path, external_storage_options=None):
 
 
 class TDEngineStoreyTarget(storey.TDEngineTarget):
-    def __init__(self, *args, **kwargs):
-        kwargs["url"] = mlrun.model_monitoring.helpers.get_tsdb_connection_string()
-        super().__init__(*args, **kwargs)
+    def __init__(self, *args, url: str, **kwargs):
+        if url.startswith("ds://"):
+            datastore_profile = (
+                mlrun.datastore.datastore_profile.datastore_profile_read(url)
+            )
+            if not isinstance(
+                datastore_profile,
+                mlrun.datastore.datastore_profile.TDEngineDatastoreProfile,
+            ):
+                raise ValueError(
+                    f"Unexpected datastore profile type:{datastore_profile.type}."
+                    "Only TDEngineDatastoreProfile is supported"
+                )
+            url = datastore_profile.dsn()
+        super().__init__(*args, url=url, **kwargs)
 
 
 class StoreyTargetUtils:
@@ -69,7 +81,12 @@ class StoreyTargetUtils:
 
 class ParquetStoreyTarget(storey.ParquetTarget):
     def __init__(self, *args, **kwargs):
+        alt_key_name = kwargs.pop("alternative_v3io_access_key", None)
         args, kwargs = StoreyTargetUtils.process_args_and_kwargs(args, kwargs)
+        storage_options = kwargs.get("storage_options", {})
+        if storage_options and storage_options.get("v3io_access_key") and alt_key_name:
+            if alt_key := mlrun.get_secret_or_env(alt_key_name):
+                storage_options["v3io_access_key"] = alt_key
         super().__init__(*args, **kwargs)
 
 

mlrun/model_monitoring/applications/base.py

@@ -95,8 +95,8 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
         sample_data: Optional[pd.DataFrame] = None,
         reference_data: Optional[pd.DataFrame] = None,
         endpoints: Optional[list[tuple[str, str]]] = None,
-        start: Optional[datetime] = None,
-        end: Optional[datetime] = None,
+        start: Optional[str] = None,
+        end: Optional[str] = None,
         base_period: Optional[int] = None,
     ):
         """
@@ -124,7 +124,6 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
             return self.do_tracking(monitoring_context)
 
         if endpoints is not None:
-            start, end = self._validate_times(start, end, base_period)
             for window_start, window_end in self._window_generator(
                 start, end, base_period
             ):
@@ -137,43 +136,40 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
                             mm_constants.ApplicationEvent.END_INFER_TIME: window_end,
                         }
                     )
-                    context.log_result(
-                        f"{endpoint_name}_{window_start.isoformat()}_{window_end.isoformat()}",
-                        result,
+                    result_key = (
+                        f"{endpoint_name}_{window_start.isoformat()}_{window_end.isoformat()}"
+                        if window_start and window_end
+                        else endpoint_name
                     )
+                    context.log_result(result_key, result)
         else:
             return call_do_tracking()
 
-    @staticmethod
-    def _validate_times(
-        start: Optional[datetime],
-        end: Optional[datetime],
-        base_period: Optional[int],
-    ) -> tuple[datetime, datetime]:
-        if (start is None) or (end is None):
-            raise mlrun.errors.MLRunValueError(
-                "When `endpoint_names` is provided, you must also pass the start and end times"
-            )
-        if (base_period is not None) and not (
-            isinstance(base_period, int) and base_period > 0
-        ):
-            raise mlrun.errors.MLRunValueError(
-                "`base_period` must be a nonnegative integer - the number of minutes in a monitoring window"
-            )
-        return start, end
-
     @staticmethod
     def _window_generator(
-        start: datetime, end: datetime, base_period: Optional[int]
-    ) -> Iterator[tuple[datetime, datetime]]:
+        start: Optional[str], end: Optional[str], base_period: Optional[int]
+    ) -> Iterator[tuple[Optional[datetime], Optional[datetime]]]:
+        if start is None or end is None:
+            # A single window based on the `sample_data` input - see `_handler`.
+            yield None, None
+            return
+
+        start_dt = datetime.fromisoformat(start)
+        end_dt = datetime.fromisoformat(end)
+
         if base_period is None:
-            yield start, end
+            yield start_dt, end_dt
             return
 
+        if not isinstance(base_period, int) or base_period <= 0:
+            raise mlrun.errors.MLRunValueError(
+                "`base_period` must be a nonnegative integer - the number of minutes in a monitoring window"
+            )
+
         window_length = timedelta(minutes=base_period)
-        current_start_time = start
-        while current_start_time < end:
-            current_end_time = min(current_start_time + window_length, end)
+        current_start_time = start_dt
+        while current_start_time < end_dt:
+            current_end_time = min(current_start_time + window_length, end_dt)
             yield current_start_time, current_end_time
             current_start_time = current_end_time
 
@@ -369,13 +365,25 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
         :param requirements:      List of Python requirements to be installed in the image.
         :param requirements_file: Path to a Python requirements file to be installed in the image.
         :param endpoints:         A list of tuples of the model endpoint (name, uid) to get the data from.
-                                  If provided, you have to provide also the start and end times of the data to analyze.
-        :param start:             The start time of the sample data.
-        :param end:               The end time of the sample data.
+                                  If provided, and ``sample_data`` is not, you have to provide also the ``start`` and
+                                  ``end`` times of the data to analyze from the model endpoints.
+        :param start:             The start time of the endpoint's data, not included.
+                                  If you want the model endpoint's data at ``start`` included, you need to subtract a
+                                  small ``datetime.timedelta`` from it.
+        :param end:               The end time of the endpoint's data, included.
+                                  Please note: when ``start`` and ``end`` are set, they create a left-open time interval
+                                  ("window") :math:`(\\text{start}, \\text{end}]` that excludes the endpoint's data at
+                                  ``start`` and includes the data at ``end``:
+                                  :math:`\\text{start} < t \\leq \\text{end}`, :math:`t` is the time taken in the
+                                  window's data.
         :param base_period:       The window length in minutes. If ``None``, the whole window from ``start`` to ``end``
                                   is taken. If an integer is specified, the application is run from ``start`` to ``end``
                                   in ``base_period`` length windows, except for the last window that ends at ``end`` and
-                                  therefore may be shorter.
+                                  therefore may be shorter:
+                                  :math:`(\\text{start}, \\text{start} + \\text{base_period}],
+                                  (\\text{start} + \\text{base_period}, \\text{start} + 2\\cdot\\text{base_period}],
+                                  ..., (\\text{start} + m\\cdot\\text{base_period}, \\text{end}]`,
+                                  where :math:`m` is some positive integer.
 
         :returns: The output of the
                   :py:meth:`~mlrun.model_monitoring.applications.ModelMonitoringApplicationBase.do_tracking`
@@ -395,16 +403,23 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
             project=project,
         )
 
-        params: dict[str, Union[list[tuple[str, str]], datetime, int, None]] = {}
+        params: dict[str, Union[list[tuple[str, str]], str, int, None]] = {}
         if endpoints:
-            start, end = cls._validate_times(start, end, base_period)
             params["endpoints"] = endpoints
-            params["start"] = start
-            params["end"] = end
-            params["base_period"] = base_period
+            if sample_data is None:
+                if start is None or end is None:
+                    raise mlrun.errors.MLRunValueError(
+                        "`start` and `end` times must be provided when `endpoints` "
+                        "is provided without `sample_data`"
+                    )
+                params["start"] = (
+                    start.isoformat() if isinstance(start, datetime) else start
+                )
+                params["end"] = end.isoformat() if isinstance(end, datetime) else end
+                params["base_period"] = base_period
         elif start or end or base_period:
             raise mlrun.errors.MLRunValueError(
-                "Custom start and end times or base_period are supported only with endpoints data"
+                "Custom `start` and `end` times or base_period are supported only with endpoints data"
             )
 
         inputs: dict[str, str] = {}

mlrun/model_monitoring/applications/results.py

@@ -33,10 +33,10 @@ class _ModelMonitoringApplicationDataRes(ABC):
     name: str
 
     def __post_init__(self):
-        pat = re.compile(r"[a-zA-Z_][a-zA-Z0-9_]*")
+        pat = re.compile(mm_constants.RESULT_NAME_PATTERN)
         if not re.fullmatch(pat, self.name):
             raise mlrun.errors.MLRunValueError(
-                "Attribute name must comply with the regex `[a-zA-Z_][a-zA-Z0-9_]*`"
+                f"Attribute name must comply with the regex `{mm_constants.RESULT_NAME_PATTERN}`"
             )
 
     @abstractmethod

mlrun/model_monitoring/controller.py

@@ -250,9 +250,10 @@ class MonitoringApplicationController:
 
         self.model_monitoring_access_key = self._get_model_monitoring_access_key()
         self.v3io_access_key = mlrun.mlconf.get_v3io_access_key()
-        self.storage_options = None
-        if mlrun.mlconf.artifact_path.startswith("s3://"):
-            self.storage_options = mlrun.mlconf.get_s3_storage_options()
+        store, _, _ = mlrun.store_manager.get_or_create_store(
+            mlrun.mlconf.artifact_path
+        )
+        self.storage_options = store.get_storage_options()
 
     @staticmethod
     def _get_model_monitoring_access_key() -> Optional[str]:

mlrun/model_monitoring/db/tsdb/__init__.py

@@ -19,6 +19,7 @@ import mlrun.common.schemas.secret
 import mlrun.datastore.datastore_profile
 import mlrun.errors
 import mlrun.model_monitoring.helpers
+from mlrun.datastore.datastore_profile import DatastoreProfile
 
 from .base import TSDBConnector
 
@@ -29,10 +30,13 @@ class ObjectTSDBFactory(enum.Enum):
     v3io_tsdb = "v3io-tsdb"
     tdengine = "tdengine"
 
-    def to_tsdb_connector(self, project: str, **kwargs) -> TSDBConnector:
+    def to_tsdb_connector(
+        self, project: str, profile: DatastoreProfile, **kwargs
+    ) -> TSDBConnector:
         """
         Return a TSDBConnector object based on the provided enum value.
         :param project: The name of the project.
+        :param profile: Datastore profile containing DSN and credentials for TSDB connection
         :return: `TSDBConnector` object.
         """
 
@@ -51,7 +55,7 @@ class ObjectTSDBFactory(enum.Enum):
 
         from .tdengine.tdengine_connector import TDEngineConnector
 
-        return TDEngineConnector(project=project, **kwargs)
+        return TDEngineConnector(project=project, profile=profile, **kwargs)
 
     @classmethod
     def _missing_(cls, value: typing.Any):
@@ -87,12 +91,10 @@ def get_tsdb_connector(
     kwargs = {}
     if isinstance(profile, mlrun.datastore.datastore_profile.DatastoreProfileV3io):
         tsdb_connector_type = mlrun.common.schemas.model_monitoring.TSDBTarget.V3IO_TSDB
-        kwargs["v3io_access_key"] = profile.v3io_access_key
     elif isinstance(
         profile, mlrun.datastore.datastore_profile.TDEngineDatastoreProfile
     ):
         tsdb_connector_type = mlrun.common.schemas.model_monitoring.TSDBTarget.TDEngine
-        kwargs["connection_string"] = profile.dsn()
     else:
         extra_message = (
             ""
@@ -109,4 +111,6 @@ def get_tsdb_connector(
     tsdb_connector_factory = ObjectTSDBFactory(tsdb_connector_type)
 
     # Convert into TSDB connector object
-    return tsdb_connector_factory.to_tsdb_connector(project=project, **kwargs)
+    return tsdb_connector_factory.to_tsdb_connector(
+        project=project, profile=profile, **kwargs
+    )