mlrun 1.10.0rc38__tar.gz → 1.10.0rc39__tar.gz

{mlrun-1.10.0rc38/mlrun.egg-info → mlrun-1.10.0rc39}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlrun
-Version: 1.10.0rc38
+Version: 1.10.0rc39
 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-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/mlrun/common/secrets.py

@@ -11,10 +11,18 @@
 # 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.
-
+import re
 from abc import ABC, abstractmethod
 
 import mlrun.common.schemas
+from mlrun.config import config as mlconf
+
+_AUTH_SECRET_NAME_TEMPLATE = re.escape(
+    mlconf.secret_stores.kubernetes.auth_secret_name.format(
+        hashed_access_key="",
+    )
+)
+AUTH_SECRET_PATTERN = re.compile(f"^{_AUTH_SECRET_NAME_TEMPLATE}.*")
 
 
 class SecretProviderInterface(ABC):

{mlrun-1.10.0rc38 → mlrun-1.10.0rc39}/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-1.10.0rc39}/mlrun/model.py

@@ -29,6 +29,7 @@ import pydantic.v1.error_wrappers
 import mlrun
 import mlrun.common.constants as mlrun_constants
 import mlrun.common.schemas.notification
+import mlrun.common.secrets
 import mlrun.utils.regex
 
 from .utils import (
@@ -1616,7 +1617,14 @@ class RunTemplate(ModelObj):
 
         :returns: The RunTemplate object
         """
-
+        if kind == "azure_vault" and isinstance(source, dict):
+            candidate_secret_name = (source.get("k8s_secret") or "").strip()
+            if candidate_secret_name and mlrun.common.secrets.AUTH_SECRET_PATTERN.match(
+                candidate_secret_name
+            ):
+                raise mlrun.errors.MLRunInvalidArgumentError(
+                    f"Forbidden secret '{candidate_secret_name}' matches MLRun auth-secret pattern."
+                )
         if kind == "vault" and isinstance(source, list):
             source = {"project": self.metadata.project, "secrets": source}
 

{mlrun-1.10.0rc38 → mlrun-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/mlrun/projects/project.py

@@ -45,6 +45,7 @@ import mlrun.common.runtimes.constants
 import mlrun.common.schemas.alert
 import mlrun.common.schemas.artifact
 import mlrun.common.schemas.model_monitoring.constants as mm_constants
+import mlrun.common.secrets
 import mlrun.datastore.datastore_profile
 import mlrun.db
 import mlrun.errors
@@ -3418,7 +3419,12 @@ class MlrunProject(ModelObj):
         self._initialized = True
         return self.spec._function_objects
 
-    def with_secrets(self, kind, source, prefix=""):
+    def with_secrets(
+        self,
+        kind,
+        source,
+        prefix="",
+    ):
         """register a secrets source (file, env or dict)
 
         read secrets from a source provider to be used in workflows, example::
@@ -3440,12 +3446,21 @@ class MlrunProject(ModelObj):
 
         This will enable access to all secrets in vault registered to the current project.
 
-        :param kind:   secret type (file, inline, env, vault)
+        :param kind:   secret type (file, inline, env, vault, azure_vault)
         :param source: secret data or link (see example)
         :param prefix: add a prefix to the keys in this source
 
         :returns: project object
         """
+        # Block using mlrun-auth-secrets.* via azure_vault's k8s_secret param (client-side only)
+        if kind == "azure_vault" and isinstance(source, dict):
+            candidate_secret_name = (source.get("k8s_secret") or "").strip()
+            if candidate_secret_name and mlrun.common.secrets.AUTH_SECRET_PATTERN.match(
+                candidate_secret_name
+            ):
+                raise mlrun.errors.MLRunInvalidArgumentError(
+                    f"Forbidden secret '{candidate_secret_name}' matches MLRun auth-secret pattern."
+                )
 
         if kind == "vault" and isinstance(source, list):
             source = {"project": self.metadata.name, "secrets": source}

{mlrun-1.10.0rc38 → mlrun-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/mlrun/runtimes/mounts.py

@@ -17,6 +17,8 @@ import typing
 import warnings
 from collections import namedtuple
 
+import mlrun.common.secrets
+import mlrun.errors
 from mlrun.config import config
 from mlrun.config import config as mlconf
 from mlrun.errors import MLRunInvalidArgumentError
@@ -412,6 +414,13 @@ def mount_secret(
                          the specified paths, and unlisted keys will not be
                          present."""
 
+    if secret_name and mlrun.common.secrets.AUTH_SECRET_PATTERN.match(
+        secret_name.strip()
+    ):
+        raise mlrun.errors.MLRunInvalidArgumentError(
+            f"Forbidden secret '{secret_name}' matches MLRun auth-secret pattern."
+        )
+
     def _mount_secret(runtime: "KubeResource"):
         # Define the secret volume source
         secret_volume_source = {

{mlrun-1.10.0rc38 → mlrun-1.10.0rc39}/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-1.10.0rc39}/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-1.10.0rc39}/mlrun/runtimes/nuclio/serving.py

@@ -44,6 +44,7 @@ from mlrun.serving.states import (
 )
 from mlrun.utils import get_caller_globals, logger, set_paths
 
+from ...common.secrets import AUTH_SECRET_PATTERN
 from .. import KubejobRuntime
 from ..pod import KubeResourceSpec
 from .function import NuclioSpec, RemoteRuntime, min_nuclio_versions
@@ -635,7 +636,14 @@ class ServingRuntime(RemoteRuntime):
 
         :returns: The Runtime (function) object
         """
-
+        if kind == "azure_vault" and isinstance(source, dict):
+            candidate_secret_name = (source.get("k8s_secret") or "").strip()
+            if candidate_secret_name and AUTH_SECRET_PATTERN.match(
+                candidate_secret_name
+            ):
+                raise mlrun.errors.MLRunInvalidArgumentError(
+                    f"Forbidden secret '{candidate_secret_name}' matches MLRun auth-secret pattern."
+                )
         if kind == "vault" and isinstance(source, list):
             source = {"project": self.metadata.project, "secrets": source}
 
@@ -659,6 +667,9 @@ 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 +866,20 @@ 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 +913,50 @@ 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-1.10.0rc39}/mlrun/runtimes/pod.py

@@ -20,6 +20,7 @@ import typing
 import warnings
 from collections.abc import Iterable
 from enum import Enum
+from typing import Optional
 
 import dotenv
 import kubernetes.client as k8s_client
@@ -34,6 +35,7 @@ from mlrun.common.schemas import (
     SecurityContextEnrichmentModes,
 )
 
+from ..common.secrets import AUTH_SECRET_PATTERN
 from ..config import config as mlconf
 from ..k8s_utils import (
     generate_preemptible_nodes_affinity_terms,
@@ -708,19 +710,44 @@ class KubeResource(BaseRuntime):
     def spec(self, spec):
         self._spec = self._verify_dict(spec, "spec", KubeResourceSpec)
 
-    def set_env_from_secret(self, name, secret=None, secret_key=None):
-        """set pod environment var from secret"""
+    def set_env_from_secret(
+        self,
+        name: str,
+        secret: Optional[str] = None,
+        secret_key: Optional[str] = None,
+    ):
+        """Set env var from secret; block auth-secret usage on client side."""
+        self._validate_no_auth_secret(
+            secret_name=secret,
+        )
         secret_key = secret_key or name
         value_from = k8s_client.V1EnvVarSource(
-            secret_key_ref=k8s_client.V1SecretKeySelector(name=secret, key=secret_key)
+            secret_key_ref=k8s_client.V1SecretKeySelector(
+                name=secret,
+                key=secret_key,
+            )
+        )
+        return self._set_env(
+            name=name,
+            value_from=value_from,
         )
-        return self._set_env(name, value_from=value_from)
 
-    def set_env(self, name, value=None, value_from=None):
-        """set pod environment var from value"""
-        if value is not None:
-            return self._set_env(name, value=str(value))
-        return self._set_env(name, value_from=value_from)
+    def set_env(
+        self,
+        name: str,
+        value: Optional[str] = None,
+        value_from: Optional[typing.Any] = None,
+    ):
+        """Set env var; block auth-secret usage when coming from a secret."""
+        if value_from is not None:
+            secret_name = self._extract_secret_name_from_value_from(
+                value_from=value_from,
+            )
+            self._validate_no_auth_secret(
+                secret_name=secret_name,
+            )
+            return self._set_env(name, value_from=value_from)
+        return self._set_env(name, value=str(value) if value is not None else None)
 
     def with_annotations(self, annotations: dict):
         """set a key/value annotations in the metadata of the pod"""
@@ -1366,6 +1393,37 @@ class KubeResource(BaseRuntime):
 
         return self.status.state
 
+    @staticmethod
+    def _validate_no_auth_secret(
+        secret_name: Optional[str],
+    ):
+        """Raise if secret name matches MLRun auth-secret pattern."""
+        if secret_name and AUTH_SECRET_PATTERN.match(secret_name):
+            raise mlrun.errors.MLRunInvalidArgumentError(
+                f"Forbidden secret '{secret_name}' matches MLRun auth-secret pattern."
+            )
+
+    @staticmethod
+    def _extract_secret_name_from_value_from(
+        value_from: typing.Any,
+    ) -> Optional[str]:
+        """Extract secret name from a V1EnvVarSource or dict representation."""
+        if isinstance(value_from, k8s_client.V1EnvVarSource):
+            if value_from.secret_key_ref:
+                return value_from.secret_key_ref.name
+        elif isinstance(value_from, dict):
+            value_from = (
+                value_from.get("valueFrom")
+                or value_from.get("value_from")
+                or value_from
+            )
+            secret_key_ref = (value_from or {}).get("secretKeyRef") or (
+                value_from or {}
+            ).get("secret_key_ref")
+            if isinstance(secret_key_ref, dict):
+                return secret_key_ref.get("name")
+        return None
+
 
 def _resolve_if_type_sanitized(attribute_name, attribute):
     attribute_config = sanitized_attributes[attribute_name]