mlrun 1.7.2rc4__py3-none-any.whl → 1.8.0__py3-none-any.whl

mlrun/common/schemas/function.py

@@ -14,7 +14,7 @@
 #
 import typing
 
-import pydantic
+import pydantic.v1
 
 import mlrun.common.types
 
@@ -90,42 +90,42 @@ class SecurityContextEnrichmentModes(mlrun.common.types.StrEnum):
     disabled = "disabled"
 
 
-class ImagePullSecret(pydantic.BaseModel):
+class ImagePullSecret(pydantic.v1.BaseModel):
     default: typing.Optional[str]
 
 
-class Pipelines(pydantic.BaseModel):
+class Pipelines(pydantic.v1.BaseModel):
     kfp_pod_user_unix_id: typing.Optional[int]
 
 
-class SecurityContext(pydantic.BaseModel):
+class SecurityContext(pydantic.v1.BaseModel):
     default: typing.Optional[str]
     enrichment_mode: typing.Optional[SecurityContextEnrichmentModes]
     enrichment_group_id: typing.Optional[int]
     pipelines: typing.Optional[Pipelines]
 
 
-class ServiceAccount(pydantic.BaseModel):
+class ServiceAccount(pydantic.v1.BaseModel):
     default: typing.Optional[str]
 
 
-class StateThresholds(pydantic.BaseModel):
+class StateThresholds(pydantic.v1.BaseModel):
     default: typing.Optional[dict[str, str]]
 
 
-class FunctionSpec(pydantic.BaseModel):
+class FunctionSpec(pydantic.v1.BaseModel):
     image_pull_secret: typing.Optional[ImagePullSecret]
     security_context: typing.Optional[SecurityContext]
     service_account: typing.Optional[ServiceAccount]
     state_thresholds: typing.Optional[StateThresholds]
 
     class Config:
-        extra = pydantic.Extra.allow
+        extra = pydantic.v1.Extra.allow
 
 
-class Function(pydantic.BaseModel):
+class Function(pydantic.v1.BaseModel):
     spec: typing.Optional[FunctionSpec]
     application: typing.Optional[dict[str, typing.Any]]
 
     class Config:
-        extra = pydantic.Extra.allow
+        extra = pydantic.v1.Extra.allow

mlrun/common/schemas/hub.py

@@ -15,7 +15,7 @@
 from datetime import datetime, timezone
 from typing import Optional
 
-from pydantic import BaseModel, Extra, Field
+from pydantic.v1 import BaseModel, Extra, Field
 
 import mlrun.common.types
 import mlrun.config

mlrun/common/schemas/k8s.py

@@ -14,18 +14,18 @@
 #
 import typing
 
-import pydantic
+import pydantic.v1
 
 import mlrun.common.types
 
 
-class ResourceSpec(pydantic.BaseModel):
+class ResourceSpec(pydantic.v1.BaseModel):
     cpu: typing.Optional[str]
     memory: typing.Optional[str]
     gpu: typing.Optional[str]
 
 
-class Resources(pydantic.BaseModel):
+class Resources(pydantic.v1.BaseModel):
     requests: ResourceSpec = ResourceSpec()
     limits: ResourceSpec = ResourceSpec()
 

mlrun/common/schemas/memory_reports.py

@@ -15,14 +15,14 @@
 
 import typing
 
-import pydantic
+import pydantic.v1
 
 
-class MostCommonObjectTypesReport(pydantic.BaseModel):
+class MostCommonObjectTypesReport(pydantic.v1.BaseModel):
     object_types: list[tuple[str, int]]
 
 
-class ObjectTypeReport(pydantic.BaseModel):
+class ObjectTypeReport(pydantic.v1.BaseModel):
     object_type: str
     sample_size: int
     start_index: typing.Optional[int]

mlrun/common/schemas/model_monitoring/__init__.py

@@ -13,8 +13,8 @@
 # limitations under the License.
 
 from .constants import (
-    V3IO_MODEL_MONITORING_DB,
-    ControllerPolicy,
+    INTERSECT_DICT_KEYS,
+    ApplicationEvent,
     DriftStatus,
     EndpointType,
     EndpointUID,
@@ -25,18 +25,16 @@ from .constants import (
     FileTargetKind,
     FunctionURI,
     MetricData,
+    ModelEndpointCreationStrategy,
     ModelEndpointMonitoringMetricType,
-    ModelEndpointTarget,
-    ModelEndpointTargetSchemas,
+    ModelEndpointSchema,
     ModelMonitoringMode,
-    ModelMonitoringStoreKinds,
     MonitoringFunctionNames,
     PredictionsQueryConstants,
     ProjectSecretKeys,
     ResultData,
     ResultKindApp,
     ResultStatusApp,
-    SchedulingKeys,
     SpecialApps,
     TDEngineSuperTables,
     TSDBTarget,
@@ -48,11 +46,9 @@ 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

@@ -29,6 +29,54 @@ class MonitoringStrEnum(StrEnum):
         return list(map(lambda c: c.value, cls))
 
 
+class ModelEndpointSchema(MonitoringStrEnum):
+    # metadata
+    UID = "uid"
+    PROJECT = "project"
+    ENDPOINT_TYPE = "endpoint_type"
+    NAME = "name"
+    CREATED = "created"
+    UPDATED = "updated"
+    LABELS = "labels"
+
+    # spec
+    FUNCTION_NAME = "function_name"
+    FUNCTION_TAG = "function_tag"
+    MODEL_NAME = "model_name"
+    MODEL_TAGS = "model_tags"
+    MODEL_PATH = "model_path"
+    MODEL_CLASS = "model_class"
+    FEATURE_NAMES = "feature_names"
+    LABEL_NAMES = "label_names"
+    FEATURE_STATS = "feature_stats"
+    MONITORING_FEATURE_SET_URI = "monitoring_feature_set_uri"
+    CHILDREN = "children"
+    CHILDREN_UIDS = "children_uids"
+    FUNCTION_URI = "function_uri"
+    MODEL_URI = "model_uri"
+
+    # status
+    STATE = "state"
+    MONITORING_MODE = "monitoring_mode"
+    FIRST_REQUEST = "first_request"
+    SAMPLING_PERCENTAGE = "sampling_percentage"
+
+    # status - operative
+    LAST_REQUEST = "last_request"
+    RESULT_STATUS = "result_status"
+    AVG_LATENCY = "avg_latency"
+    ERROR_COUNT = "error_count"
+    CURRENT_STATS = "current_stats"
+    DRIFT_MEASURES = "drift_measures"
+
+
+class ModelEndpointCreationStrategy(MonitoringStrEnum):
+    INPLACE = "inplace"
+    ARCHIVE = "archive"
+    OVERWRITE = "overwrite"
+    SKIP = "skip"
+
+
 class EventFieldType:
     FUNCTION_URI = "function_uri"
     FUNCTION = "function"
@@ -40,6 +88,7 @@ class EventFieldType:
     TIMESTAMP = "timestamp"
     # `endpoint_id` is deprecated as a field in the model endpoint schema since 1.3.1, replaced by `uid`.
     ENDPOINT_ID = "endpoint_id"
+    ENDPOINT_NAME = "endpoint_name"
     UID = "uid"
     ENDPOINT_TYPE = "endpoint_type"
     REQUEST_ID = "request_id"
@@ -55,6 +104,8 @@ class EventFieldType:
     NAMED_PREDICTIONS = "named_predictions"
     ERROR_COUNT = "error_count"
     MODEL_ERROR = "model_error"
+    ERROR_TYPE = "error_type"
+    INFER_ERROR = "infer_error"
     ENTITIES = "entities"
     FIRST_REQUEST = "first_request"
     LAST_REQUEST = "last_request"
@@ -85,11 +136,14 @@ class EventFieldType:
     SAMPLE_PARQUET_PATH = "sample_parquet_path"
     TIME = "time"
     TABLE_COLUMN = "table_column"
+    SAMPLING_PERCENTAGE = "sampling_percentage"
+    SAMPLING_RATE = "sampling_rate"
+    ESTIMATED_PREDICTION_COUNT = "estimated_prediction_count"
+    EFFECTIVE_SAMPLE_COUNT = "effective_sample_count"
 
 
 class FeatureSetFeatures(MonitoringStrEnum):
     LATENCY = EventFieldType.LATENCY
-    ERROR_COUNT = EventFieldType.ERROR_COUNT
     METRICS = EventFieldType.METRICS
 
     @classmethod
@@ -106,21 +160,49 @@ class ApplicationEvent:
     START_INFER_TIME = "start_infer_time"
     END_INFER_TIME = "end_infer_time"
     ENDPOINT_ID = "endpoint_id"
-    OUTPUT_STREAM_URI = "output_stream_uri"
+    ENDPOINT_NAME = "endpoint_name"
+    ENDPOINT_UPDATED = "endpoint_updated"
 
 
 class WriterEvent(MonitoringStrEnum):
+    ENDPOINT_NAME = "endpoint_name"
     APPLICATION_NAME = "application_name"
     ENDPOINT_ID = "endpoint_id"
     START_INFER_TIME = "start_infer_time"
     END_INFER_TIME = "end_infer_time"
-    EVENT_KIND = "event_kind"  # metric or result
+    EVENT_KIND = "event_kind"  # metric or result or stats
     DATA = "data"
 
 
 class WriterEventKind(MonitoringStrEnum):
     METRIC = "metric"
     RESULT = "result"
+    STATS = "stats"
+
+
+class ControllerEvent(MonitoringStrEnum):
+    KIND = "kind"
+    ENDPOINT_ID = "endpoint_id"
+    ENDPOINT_NAME = "endpoint_name"
+    PROJECT = "project"
+    TIMESTAMP = "timestamp"
+    FIRST_REQUEST = "first_request"
+    FEATURE_SET_URI = "feature_set_uri"
+    ENDPOINT_TYPE = "endpoint_type"
+    ENDPOINT_POLICY = "endpoint_policy"
+    # Note: currently under endpoint policy we will have a dictionary including the keys: "application_names"
+    # "base_period", and "updated_endpoint" stand for when the MEP was updated
+
+
+class ControllerEventEndpointPolicy(MonitoringStrEnum):
+    BASE_PERIOD = "base_period"
+    MONITORING_APPLICATIONS = "monitoring_applications"
+    ENDPOINT_UPDATED = "endpoint_updated"
+
+
+class ControllerEventKind(MonitoringStrEnum):
+    NOP_EVENT = "nop_event"
+    REGULAR_EVENT = "regular_event"
 
 
 class MetricData(MonitoringStrEnum):
@@ -134,7 +216,17 @@ class ResultData(MonitoringStrEnum):
     RESULT_KIND = "result_kind"
     RESULT_STATUS = "result_status"
     RESULT_EXTRA_DATA = "result_extra_data"
+
+
+class StatsData(MonitoringStrEnum):
+    STATS_NAME = "stats_name"
+    STATS = "stats"
+    TIMESTAMP = "timestamp"
+
+
+class StatsKind(MonitoringStrEnum):
     CURRENT_STATS = "current_stats"
+    DRIFT_MEASURES = "drift_measures"
 
 
 class EventLiveStats:
@@ -153,52 +245,28 @@ class EventKeyMetrics:
     REAL_TIME = "real_time"
 
 
-class ModelEndpointTarget(MonitoringStrEnum):
-    V3IO_NOSQL = "v3io-nosql"
-    SQL = "sql"
-
-
-class StreamKind(MonitoringStrEnum):
-    V3IO_STREAM = "v3io_stream"
-    KAFKA = "kafka"
-
-
 class TSDBTarget(MonitoringStrEnum):
     V3IO_TSDB = "v3io-tsdb"
     TDEngine = "tdengine"
 
 
 class ProjectSecretKeys:
-    ENDPOINT_STORE_CONNECTION = "MODEL_MONITORING_ENDPOINT_STORE_CONNECTION"
     ACCESS_KEY = "MODEL_MONITORING_ACCESS_KEY"
-    STREAM_PATH = "STREAM_PATH"
-    TSDB_CONNECTION = "TSDB_CONNECTION"
+    TSDB_PROFILE_NAME = "TSDB_PROFILE_NAME"
+    STREAM_PROFILE_NAME = "STREAM_PROFILE_NAME"
 
     @classmethod
     def mandatory_secrets(cls):
         return [
-            cls.ENDPOINT_STORE_CONNECTION,
-            cls.STREAM_PATH,
-            cls.TSDB_CONNECTION,
+            cls.STREAM_PROFILE_NAME,
+            cls.TSDB_PROFILE_NAME,
         ]
 
 
-class ModelEndpointTargetSchemas(MonitoringStrEnum):
-    V3IO = "v3io"
-    MYSQL = "mysql"
-    SQLITE = "sqlite"
-
-
-class ModelMonitoringStoreKinds:
-    ENDPOINTS = "endpoints"
-    EVENTS = "events"
-
-
-class SchedulingKeys:
-    LAST_ANALYZED = "last_analyzed"
-    ENDPOINT_ID = "endpoint_id"
-    APPLICATION_NAME = "application_name"
-    UID = "uid"
+class GetEventsFormat(MonitoringStrEnum):
+    SINGLE = "single"
+    SEPARATION = "separation"
+    INTERSECTION = "intersection"
 
 
 class FileTargetKind:
@@ -209,22 +277,32 @@ class FileTargetKind:
     PARQUET = "parquet"
     APPS_PARQUET = "apps_parquet"
     LOG_STREAM = "log_stream"
-    APP_RESULTS = "app_results"
-    APP_METRICS = "app_metrics"
     MONITORING_SCHEDULES = "monitoring_schedules"
     MONITORING_APPLICATION = "monitoring_application"
     ERRORS = "errors"
+    STATS = "stats"
+    LAST_REQUEST = "last_request"
 
 
-class ModelMonitoringMode(str, Enum):
+class ModelMonitoringMode(StrEnum):
     enabled = "enabled"
     disabled = "disabled"
 
 
+class ScheduleChiefFields(StrEnum):
+    LAST_REQUEST = "last_request"
+    LAST_ANALYZED = "last_analyzed"
+
+
 class EndpointType(IntEnum):
     NODE_EP = 1  # end point that is not a child of a router
     ROUTER = 2  # endpoint that is router
     LEAF_EP = 3  # end point that is a child of a router
+    BATCH_EP = 4  # endpoint that is representing an offline batch endpoint
+
+    @classmethod
+    def top_level_list(cls):
+        return [cls.NODE_EP, cls.ROUTER, cls.BATCH_EP]
 
 
 class MonitoringFunctionNames(MonitoringStrEnum):
@@ -238,12 +316,14 @@ class V3IOTSDBTables(MonitoringStrEnum):
     METRICS = "metrics"
     EVENTS = "events"
     ERRORS = "errors"
+    PREDICTIONS = "predictions"
 
 
 class TDEngineSuperTables(MonitoringStrEnum):
     APP_RESULTS = "app_results"
     METRICS = "metrics"
     PREDICTIONS = "predictions"
+    ERRORS = "errors"
 
 
 @dataclass
@@ -344,10 +424,6 @@ class ModelMonitoringAppLabel:
         return f"{self.KEY}={self.VAL}"
 
 
-class ControllerPolicy:
-    BASE_PERIOD = "base_period"
-
-
 class HistogramDataDriftApplicationConstants:
     NAME = "histogram-data-drift"
     GENERAL_RESULT_NAME = "general_drift"
@@ -365,9 +441,6 @@ class SpecialApps:
 _RESERVED_FUNCTION_NAMES = MonitoringFunctionNames.list() + [SpecialApps.MLRUN_INFRA]
 
 
-V3IO_MODEL_MONITORING_DB = "v3io"
-
-
 class ModelEndpointMonitoringMetricType(StrEnum):
     RESULT = "result"
     METRIC = "metric"
@@ -384,5 +457,13 @@ 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",
+    ModelEndpointMonitoringMetricType.RESULT: "intersect_results",
+}
+
+CRON_TRIGGER_KINDS = ("http", "cron")
+STREAM_TRIGGER_KINDS = ("v3io-stream", "kafka-cluster")

mlrun/common/schemas/model_monitoring/grafana.py

@@ -14,7 +14,7 @@
 
 from typing import Optional, Union
 
-from pydantic import BaseModel
+from pydantic.v1 import BaseModel
 
 import mlrun.common.types
 
@@ -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),
+        ]