mlrun 1.10.0rc38__tar.gz → 1.10.0rc40__tar.gz

{mlrun-1.10.0rc38/mlrun.egg-info → mlrun-1.10.0rc40}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlrun
-Version: 1.10.0rc38
+Version: 1.10.0rc40
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
@@ -113,7 +113,7 @@ Requires-Dist: apscheduler<4,>=3.11; extra == "api"
 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~=0.116.0; extra == "api"
+Requires-Dist: fastapi~=0.120.0; 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"
@@ -203,7 +203,7 @@ Requires-Dist: dask~=2023.12.1; python_version < "3.11" and extra == "complete-a
 Requires-Dist: databricks-sdk~=0.20.0; extra == "complete-api"
 Requires-Dist: distributed==2024.8; python_version >= "3.11" and extra == "complete-api"
 Requires-Dist: distributed~=2023.12.1; python_version < "3.11" and extra == "complete-api"
-Requires-Dist: fastapi~=0.116.0; extra == "complete-api"
+Requires-Dist: fastapi~=0.120.0; extra == "complete-api"
 Requires-Dist: gcsfs<=2025.7.0,>=2025.5.1; extra == "complete-api"
 Requires-Dist: google-cloud-bigquery-storage~=2.17; extra == "complete-api"
 Requires-Dist: google-cloud-bigquery[bqstorage,pandas]==3.14.1; extra == "complete-api"

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/dockerfiles/mlrun-api/requirements.txt

@@ -6,7 +6,7 @@ apscheduler>=3.11,<4
 objgraph~=3.6
 igz-mgmt~=0.4.1
 humanfriendly~=10.0
-fastapi~=0.116.0
+fastapi~=0.120.0
 sqlalchemy~=2.0
 sqlalchemy-utils~=0.41.2
 pymysql~=1.1

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/artifacts/document.py

@@ -359,7 +359,12 @@ class DocumentArtifact(Artifact):
         self,
         splitter: Optional["TextSplitter"] = None,  # noqa: F821
     ) -> list["Document"]:  # noqa: F821
-        from langchain.schema import Document
+        # Try new langchain 1.0+ import path first
+        try:
+            from langchain_core.documents import Document
+        except ImportError:
+            # Fall back to old langchain <1.0 import path
+            from langchain.schema import Document
 
         """
         Create LC documents from the artifact

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/common/constants.py

@@ -27,6 +27,12 @@ DASK_LABEL_PREFIX = "dask.org/"
 NUCLIO_LABEL_PREFIX = "nuclio.io/"
 RESERVED_TAG_NAME_LATEST = "latest"
 
+# Kubernetes DNS-1123 label name length limit
+K8S_DNS_1123_LABEL_MAX_LENGTH = 63
+
+
+RESERVED_BATCH_JOB_SUFFIX = "-batch"
+
 JOB_TYPE_WORKFLOW_RUNNER = "workflow-runner"
 JOB_TYPE_PROJECT_LOADER = "project-loader"
 JOB_TYPE_RERUN_WORKFLOW_RUNNER = "rerun-workflow-runner"

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/common/schemas/model_monitoring/constants.py

@@ -486,8 +486,6 @@ class ModelMonitoringLabels:
 
 _RESERVED_FUNCTION_NAMES = MonitoringFunctionNames.list() + [SpecialApps.MLRUN_INFRA]
 
-_RESERVED_EVALUATE_FUNCTION_SUFFIX = "-batch"
-
 
 class ModelEndpointMonitoringMetricType(StrEnum):
     RESULT = "result"

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/launcher/local.py

@@ -243,6 +243,8 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
 
         # if the handler has module prefix force "local" (vs "handler") runtime
         kind = "local" if isinstance(handler, str) and "." in handler else ""
+
+        # Create temporary local function for execution
         fn = mlrun.new_function(meta.name, command=command, args=args, kind=kind)
         fn.metadata = meta
         setattr(fn, "_is_run_local", True)

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/model_monitoring/api.py

@@ -563,9 +563,10 @@ def _create_model_monitoring_function_base(
             "An application cannot have the following names: "
             f"{mm_constants._RESERVED_FUNCTION_NAMES}"
         )
-    if name and name.endswith(mm_constants._RESERVED_EVALUATE_FUNCTION_SUFFIX):
+    _, has_valid_suffix, suffix = mlrun.utils.helpers.ensure_batch_job_suffix(name)
+    if name and not has_valid_suffix:
         raise mlrun.errors.MLRunValueError(
-            "Model monitoring application names cannot end with `-batch`"
+            f"Model monitoring application names cannot end with `{suffix}`"
         )
     if func is None:
         func = ""

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/model_monitoring/applications/base.py

@@ -799,10 +799,13 @@ class ModelMonitoringApplicationBase(MonitoringApplicationToDict, ABC):
                 f"`{mm_constants.APP_NAME_REGEX.pattern}`. "
                 "Please choose another `func_name`."
             )
-        if not job_name.endswith(mm_constants._RESERVED_EVALUATE_FUNCTION_SUFFIX):
-            job_name += mm_constants._RESERVED_EVALUATE_FUNCTION_SUFFIX
+        job_name, was_renamed, suffix = mlrun.utils.helpers.ensure_batch_job_suffix(
+            job_name
+        )
+        if was_renamed:
             mlrun.utils.logger.info(
-                'Changing function name - adding `"-batch"` suffix', func_name=job_name
+                f'Changing function name - adding `"{suffix}"` suffix',
+                func_name=job_name,
             )
 
         return job_name

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/run.py

@@ -555,6 +555,7 @@ def new_function(
 
     # make sure function name is valid
     name = mlrun.utils.helpers.normalize_name(name)
+    mlrun.utils.helpers.validate_function_name(name)
 
     runner.metadata.name = name
     runner.metadata.project = (
@@ -594,6 +595,7 @@ def new_function(
         )
 
     runner.prepare_image_for_deploy()
+
     return runner
 
 
@@ -798,6 +800,9 @@ def code_to_function(
         kind=sub_kind,
         ignored_tags=ignored_tags,
     )
+
+    mlrun.utils.helpers.validate_function_name(name)
+
     spec["spec"]["env"].append(
         {
             "name": "MLRUN_HTTPDB__NUCLIO__EXPLICIT_ACK",
@@ -850,6 +855,7 @@ def code_to_function(
         runtime.spec.build.code_origin = code_origin
         runtime.spec.build.origin_filename = filename or (name + ".ipynb")
         update_common(runtime, spec)
+
         return runtime
 
     if kind is None or kind in ["", "Function"]:
@@ -863,6 +869,7 @@ def code_to_function(
 
     if not name:
         raise ValueError("name must be specified")
+
     h = get_in(spec, "spec.handler", "").split(":")
     runtime.handler = h[0] if len(h) <= 1 else h[1]
     runtime.metadata = get_in(spec, "spec.metadata")

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/__init__.py

@@ -221,6 +221,24 @@ class RuntimeKinds:
             return True
         return False
 
+    @staticmethod
+    def requires_k8s_name_validation(kind: str) -> bool:
+        """
+        Returns True if the runtime kind creates Kubernetes resources that use the function name.
+
+        Function names for k8s-deployed runtimes must conform to DNS-1123 label requirements:
+        - Lowercase alphanumeric characters or '-'
+        - Start and end with an alphanumeric character
+        - Maximum 63 characters
+
+        Local runtimes (local, handler) run on the local machine and don't create k8s resources,
+        so they don't require k8s naming validation.
+
+        :param kind: Runtime kind string (job, spark, serving, local, etc.)
+        :return: True if function name needs k8s DNS-1123 validation, False otherwise
+        """
+        return not RuntimeKinds.is_local_runtime(kind)
+
     @staticmethod
     def requires_absolute_artifacts_path(kind):
         """

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/base.py

@@ -393,6 +393,9 @@ class BaseRuntime(ModelObj):
                 FutureWarning,
             )
         output_path = output_path or out_path or artifact_path
+
+        mlrun.utils.helpers.validate_function_name(self.metadata.name)
+
         launcher = mlrun.launcher.factory.LauncherFactory().create_launcher(
             self._is_remote, local=local, **launcher_kwargs
         )

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/local.py

@@ -29,12 +29,12 @@ from os import environ, remove
 from pathlib import Path
 from subprocess import PIPE, Popen
 from sys import executable
+from typing import Optional
 
 from nuclio import Event
 
 import mlrun
 import mlrun.common.constants as mlrun_constants
-import mlrun.common.runtimes.constants
 from mlrun.lists import RunList
 
 from ..errors import err_to_str
@@ -201,9 +201,12 @@ class LocalRuntime(BaseRuntime, ParallelRunner):
     kind = "local"
     _is_remote = False
 
-    def to_job(self, image=""):
+    def to_job(self, image="", func_name: Optional[str] = None):
         struct = self.to_dict()
         obj = KubejobRuntime.from_dict(struct)
+        obj.kind = "job"  # Ensure kind is set to 'job' for KubejobRuntime
+        if func_name:
+            obj.metadata.name = func_name
         if image:
             obj.spec.image = image
         return obj

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/nuclio/application/application.py

@@ -400,6 +400,8 @@ class ApplicationRuntime(RemoteRuntime):
 
         :return: The default API gateway URL if created or True if the function is ready (deployed)
         """
+        mlrun.utils.helpers.validate_function_name(self.metadata.name)
+
         if (self.requires_build() and not self.spec.image) or force_build:
             self._fill_credentials()
             self._build_application_image(

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/nuclio/function.py

@@ -655,6 +655,8 @@ class RemoteRuntime(KubeResource):
         if tag:
             self.metadata.tag = tag
 
+        mlrun.utils.helpers.validate_function_name(self.metadata.name)
+
         # Attempt auto-mounting, before sending to remote build
         self.try_auto_mount_based_on_config()
         self._fill_credentials()

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/runtimes/nuclio/serving.py

@@ -659,6 +659,9 @@ class ServingRuntime(RemoteRuntime):
         :param builder_env: env vars dict for source archive config/credentials e.g. builder_env={"GIT_TOKEN": token}
         :param force_build: set True for force building the image
         """
+        # Validate function name before deploying to k8s
+        mlrun.utils.helpers.validate_function_name(self.metadata.name)
+
         load_mode = self.spec.load_mode
         if load_mode and load_mode not in ["sync", "async"]:
             raise ValueError(f"illegal model loading mode {load_mode}")
@@ -855,8 +858,20 @@ class ServingRuntime(RemoteRuntime):
         )
         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."""
+    def to_job(self, func_name: Optional[str] = None) -> KubejobRuntime:
+        """Convert this ServingRuntime to a KubejobRuntime, so that the graph can be run as a standalone job.
+
+        Args:
+            func_name: Optional custom name for the job function. If not provided, automatically
+                      appends '-batch' suffix to the serving function name to prevent database collision.
+
+        Returns:
+            KubejobRuntime configured to execute the serving graph as a batch job.
+
+        Note:
+            The job will have a different name than the serving function to prevent database collision.
+            The original serving function remains unchanged and can still be invoked after running the 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"
@@ -890,8 +905,50 @@ class ServingRuntime(RemoteRuntime):
             parameters=self.spec.parameters,
             graph=self.spec.graph,
         )
+
+        job_metadata = deepcopy(self.metadata)
+        original_name = job_metadata.name
+
+        if func_name:
+            # User provided explicit job name
+            job_metadata.name = func_name
+            logger.debug(
+                "Creating job from serving function with custom name",
+                new_name=func_name,
+            )
+        else:
+            job_metadata.name, was_renamed, suffix = (
+                mlrun.utils.helpers.ensure_batch_job_suffix(job_metadata.name)
+            )
+
+            # Check if the resulting name exceeds Kubernetes length limit
+            if (
+                len(job_metadata.name)
+                > mlrun.common.constants.K8S_DNS_1123_LABEL_MAX_LENGTH
+            ):
+                raise mlrun.errors.MLRunInvalidArgumentError(
+                    f"Cannot convert serving function '{original_name}' to batch job: "
+                    f"the resulting name '{job_metadata.name}' ({len(job_metadata.name)} characters) "
+                    f"exceeds Kubernetes limit of {mlrun.common.constants.K8S_DNS_1123_LABEL_MAX_LENGTH} characters. "
+                    f"Please provide a custom name via the func_name parameter, "
+                    f"with at most {mlrun.common.constants.K8S_DNS_1123_LABEL_MAX_LENGTH} characters."
+                )
+
+            if was_renamed:
+                logger.info(
+                    "Creating job from serving function (auto-appended suffix to prevent collision)",
+                    new_name=job_metadata.name,
+                    suffix=suffix,
+                )
+            else:
+                logger.debug(
+                    "Creating job from serving function (name already has suffix)",
+                    name=original_name,
+                    suffix=suffix,
+                )
+
         job = KubejobRuntime(
             spec=spec,
-            metadata=self.metadata,
+            metadata=job_metadata,
         )
         return job

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun/utils/helpers.py

@@ -253,6 +253,40 @@ def verify_field_regex(
         return False
 
 
+def validate_function_name(name: str) -> None:
+    """
+    Validate that a function name conforms to Kubernetes DNS-1123 label requirements.
+
+    Function names for Kubernetes resources must:
+    - Be lowercase alphanumeric characters or '-'
+    - Start and end with an alphanumeric character
+    - Be at most 63 characters long
+
+    This validation should be called AFTER normalize_name() has been applied.
+
+    Refer to https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#dns-label-names
+
+    :param name: The function name to validate (after normalization)
+    :raises MLRunInvalidArgumentError: If the function name is invalid for Kubernetes
+    """
+    if not name:
+        return
+
+    verify_field_regex(
+        "function.metadata.name",
+        name,
+        mlrun.utils.regex.dns_1123_label,
+        raise_on_failure=True,
+        log_message=(
+            f"Function name '{name}' is invalid. "
+            "Kubernetes function names must be DNS-1123 labels: "
+            "lowercase alphanumeric characters or '-', "
+            "starting and ending with an alphanumeric character, "
+            "and at most 63 characters long."
+        ),
+    )
+
+
 def validate_builder_source(
     source: str, pull_at_runtime: bool = False, workdir: Optional[str] = None
 ):
@@ -476,6 +510,40 @@ def normalize_name(name: str):
     return name.lower()
 
 
+def ensure_batch_job_suffix(
+    function_name: typing.Optional[str],
+) -> tuple[typing.Optional[str], bool, str]:
+    """
+    Ensure that a function name has the batch job suffix appended to prevent database collision.
+
+    This helper is used by to_job() methods in runtimes that convert online functions (serving, local)
+    to batch processing jobs. The suffix prevents the job from overwriting the original function in
+    the database when both are stored with the same (project, name) key.
+
+    :param function_name: The original function name (can be None or empty string)
+
+    :return: A tuple of (modified_name, was_renamed, suffix) where:
+        - modified_name: The function name with the batch suffix (if not already present),
+          or empty string if input was empty
+        - was_renamed: True if the suffix was added, False if it was already present or if name was empty/None
+        - suffix: The suffix value that was used (or would have been used)
+
+    """
+    suffix = mlrun_constants.RESERVED_BATCH_JOB_SUFFIX
+
+    # Handle None or empty string
+    if not function_name:
+        return function_name, False, suffix
+
+    if not function_name.endswith(suffix):
+        return (
+            f"{function_name}{suffix}",
+            True,
+            suffix,
+        )
+    return function_name, False, suffix
+
+
 class LogBatchWriter:
     def __init__(self, func, batch=16, maxtime=5):
         self.batch = batch
@@ -970,8 +1038,15 @@ def enrich_image_url(
         else:
             image_url = "mlrun/mlrun"
 
-    if is_mlrun_image and tag and ":" not in image_url:
-        image_url = f"{image_url}:{tag}"
+    if is_mlrun_image and tag:
+        if ":" not in image_url:
+            image_url = f"{image_url}:{tag}"
+        elif enrich_kfp_python_version:
+            # For mlrun-kfp >= 1.10.0-rc0, append python suffix to existing tag
+            python_suffix = resolve_image_tag_suffix(
+                mlrun_version, client_python_version
+            )
+            image_url = f"{image_url}{python_suffix}" if python_suffix else image_url
 
     registry = (
         config.images_registry if is_mlrun_image else config.vendor_images_registry

mlrun-1.10.0rc40/mlrun/utils/version/version.json

@@ -0,0 +1,4 @@
+{
+  "git_commit": "3bb8c631304e89739430fbb6e6299e190fe6738a",
+  "version": "1.10.0-rc40"
+}

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40/mlrun.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlrun
-Version: 1.10.0rc38
+Version: 1.10.0rc40
 Summary: Tracking and config of machine learning runs
 Home-page: https://github.com/mlrun/mlrun
 Author: Yaron Haviv
@@ -113,7 +113,7 @@ Requires-Dist: apscheduler<4,>=3.11; extra == "api"
 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~=0.116.0; extra == "api"
+Requires-Dist: fastapi~=0.120.0; 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"
@@ -203,7 +203,7 @@ Requires-Dist: dask~=2023.12.1; python_version < "3.11" and extra == "complete-a
 Requires-Dist: databricks-sdk~=0.20.0; extra == "complete-api"
 Requires-Dist: distributed==2024.8; python_version >= "3.11" and extra == "complete-api"
 Requires-Dist: distributed~=2023.12.1; python_version < "3.11" and extra == "complete-api"
-Requires-Dist: fastapi~=0.116.0; extra == "complete-api"
+Requires-Dist: fastapi~=0.120.0; extra == "complete-api"
 Requires-Dist: gcsfs<=2025.7.0,>=2025.5.1; extra == "complete-api"
 Requires-Dist: google-cloud-bigquery-storage~=2.17; extra == "complete-api"
 Requires-Dist: google-cloud-bigquery[bqstorage,pandas]==3.14.1; extra == "complete-api"

{mlrun-1.10.0rc38 → mlrun-1.10.0rc40}/mlrun.egg-info/requires.txt

@@ -85,7 +85,7 @@ apscheduler<4,>=3.11
 objgraph~=3.6
 igz-mgmt~=0.4.1
 humanfriendly~=10.0
-fastapi~=0.116.0
+fastapi~=0.120.0
 sqlalchemy~=2.0
 sqlalchemy-utils~=0.41.2
 pymysql~=1.1
@@ -151,7 +151,7 @@ azure-keyvault-secrets~=4.2
 boto3<1.36,>=1.28.0
 dask-kubernetes~=0.11.0
 databricks-sdk~=0.20.0
-fastapi~=0.116.0
+fastapi~=0.120.0
 gcsfs<=2025.7.0,>=2025.5.1
 google-cloud-bigquery-storage~=2.17
 google-cloud-bigquery[bqstorage,pandas]==3.14.1

mlrun-1.10.0rc38/mlrun/utils/version/version.json

@@ -1,4 +0,0 @@
-{
-  "git_commit": "cc5c5639d721f37d6a1d0d0b7cf9f853f38e4707",
-  "version": "1.10.0-rc38"
-}