apache-airflow-providers-cncf-kubernetes 7.11.0__tar.gz → 7.12.0__tar.gz

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: apache-airflow-providers-cncf-kubernetes
-Version: 7.11.0
+Version: 7.12.0
 Summary: Provider package apache-airflow-providers-cncf-kubernetes for Apache Airflow
 Keywords: airflow-provider,cncf.kubernetes,airflow,integration
 Author-email: Apache Software Foundation <dev@airflow.apache.org>
@@ -28,8 +28,8 @@ Requires-Dist: google-re2>=1.0
 Requires-Dist: kubernetes>=21.7.0,<24
 Requires-Dist: kubernetes_asyncio>=18.20.1,<25
 Project-URL: Bug Tracker, https://github.com/apache/airflow/issues
-Project-URL: Changelog, https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/changelog.html
-Project-URL: Documentation, https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0
+Project-URL: Changelog, https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/changelog.html
+Project-URL: Documentation, https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0
 Project-URL: Slack Chat, https://s.apache.org/airflow-slack
 Project-URL: Source Code, https://github.com/apache/airflow
 Project-URL: Twitter, https://twitter.com/ApacheAirflow
@@ -79,7 +79,7 @@ Project-URL: YouTube, https://www.youtube.com/channel/UCSXwxpWZQ7XZ1WL3wqevChA/
 
 Package ``apache-airflow-providers-cncf-kubernetes``
 
-Release: ``7.11.0``
+Release: ``7.12.0``
 
 
 `Kubernetes <https://kubernetes.io/>`__
@@ -92,7 +92,7 @@ This is a provider package for ``cncf.kubernetes`` provider. All classes for thi
 are in ``airflow.providers.cncf.kubernetes`` python package.
 
 You can find package information and changelog for the provider
-in the `documentation <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/>`_.
+in the `documentation <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/>`_.
 
 Installation
 ------------
@@ -119,4 +119,4 @@ PIP package             Version required
 ======================  ==================
 
 The changelog for the provider package can be found in the
-`changelog <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/changelog.html>`_.
+`changelog <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/changelog.html>`_.

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/README.rst

@@ -42,7 +42,7 @@
 
 Package ``apache-airflow-providers-cncf-kubernetes``
 
-Release: ``7.11.0``
+Release: ``7.12.0``
 
 
 `Kubernetes <https://kubernetes.io/>`__
@@ -55,7 +55,7 @@ This is a provider package for ``cncf.kubernetes`` provider. All classes for thi
 are in ``airflow.providers.cncf.kubernetes`` python package.
 
 You can find package information and changelog for the provider
-in the `documentation <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/>`_.
+in the `documentation <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/>`_.
 
 Installation
 ------------
@@ -82,4 +82,4 @@ PIP package             Version required
 ======================  ==================
 
 The changelog for the provider package can be found in the
-`changelog <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/changelog.html>`_.
+`changelog <https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/changelog.html>`_.

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/__init__.py

@@ -27,7 +27,7 @@ import packaging.version
 
 __all__ = ["__version__"]
 
-__version__ = "7.11.0"
+__version__ = "7.12.0"
 
 try:
     from airflow import __version__ as airflow_version

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/executors/kubernetes_executor.py

@@ -34,6 +34,7 @@ from datetime import datetime
 from queue import Empty, Queue
 from typing import TYPE_CHECKING, Any, Sequence
 
+from kubernetes.dynamic import DynamicClient
 from sqlalchemy import select, update
 
 from airflow.providers.cncf.kubernetes.pod_generator import PodMutationHookException, PodReconciliationError
@@ -74,7 +75,10 @@ except ImportError:
     raise
 from airflow.configuration import conf
 from airflow.executors.base_executor import BaseExecutor
-from airflow.providers.cncf.kubernetes.executors.kubernetes_executor_types import POD_EXECUTOR_DONE_KEY
+from airflow.providers.cncf.kubernetes.executors.kubernetes_executor_types import (
+    ADOPTED,
+    POD_EXECUTOR_DONE_KEY,
+)
 from airflow.providers.cncf.kubernetes.kube_config import KubeConfig
 from airflow.providers.cncf.kubernetes.kubernetes_helper_functions import annotations_to_key
 from airflow.utils.event_scheduler import EventScheduler
@@ -160,19 +164,31 @@ class KubernetesExecutor(BaseExecutor):
         super().__init__(parallelism=self.kube_config.parallelism)
 
     def _list_pods(self, query_kwargs):
+        query_kwargs["header_params"] = {
+            "Accept": "application/json;as=PartialObjectMetadataList;v=v1;g=meta.k8s.io"
+        }
+        dynamic_client = DynamicClient(self.kube_client.api_client)
+        pod_resource = dynamic_client.resources.get(api_version="v1", kind="Pod")
         if self.kube_config.multi_namespace_mode:
             if self.kube_config.multi_namespace_mode_namespace_list:
-                pods = []
-                for namespace in self.kube_config.multi_namespace_mode_namespace_list:
-                    pods.extend(
-                        self.kube_client.list_namespaced_pod(namespace=namespace, **query_kwargs).items
-                    )
+                namespaces = self.kube_config.multi_namespace_mode_namespace_list
             else:
-                pods = self.kube_client.list_pod_for_all_namespaces(**query_kwargs).items
+                namespaces = [None]
         else:
-            pods = self.kube_client.list_namespaced_pod(
-                namespace=self.kube_config.kube_namespace, **query_kwargs
-            ).items
+            namespaces = [self.kube_config.kube_namespace]
+
+        pods = []
+        for namespace in namespaces:
+            # Dynamic Client list pods is throwing TypeError when there are no matching pods to return
+            # This bug was fixed in MR https://github.com/kubernetes-client/python/pull/2155
+            # TODO: Remove the try-except clause once we upgrade the K8 Python client version which
+            # includes the above MR
+            try:
+                pods.extend(
+                    dynamic_client.get(resource=pod_resource, namespace=namespace, **query_kwargs).items
+                )
+            except TypeError:
+                continue
 
         return pods
 
@@ -450,7 +466,7 @@ class KubernetesExecutor(BaseExecutor):
     def _change_state(
         self,
         key: TaskInstanceKey,
-        state: TaskInstanceState | None,
+        state: TaskInstanceState | str | None,
         pod_name: str,
         namespace: str,
         session: Session = NEW_SESSION,
@@ -458,6 +474,15 @@ class KubernetesExecutor(BaseExecutor):
         if TYPE_CHECKING:
             assert self.kube_scheduler
 
+        if state == ADOPTED:
+            # When the task pod is adopted by another executor,
+            # then remove the task from the current executor running queue.
+            try:
+                self.running.remove(key)
+            except KeyError:
+                self.log.debug("TI key not in running: %s", key)
+            return
+
         if state == TaskInstanceState.RUNNING:
             self.event_buffer[key] = state, None
             return

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/executors/kubernetes_executor_types.py

@@ -16,8 +16,9 @@
 # under the License.
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, Any, Dict, Optional, Tuple
+from typing import TYPE_CHECKING, Any, Dict, Optional, Tuple, Union
 
+ADOPTED = "adopted"
 if TYPE_CHECKING:
     from airflow.executors.base_executor import CommandType
     from airflow.models.taskinstance import TaskInstanceKey
@@ -27,10 +28,10 @@ if TYPE_CHECKING:
     KubernetesJobType = Tuple[TaskInstanceKey, CommandType, Any, Optional[str]]
 
     # key, pod state, pod_name, namespace, resource_version
-    KubernetesResultsType = Tuple[TaskInstanceKey, Optional[TaskInstanceState], str, str, str]
+    KubernetesResultsType = Tuple[TaskInstanceKey, Optional[Union[TaskInstanceState, str]], str, str, str]
 
     # pod_name, namespace, pod state, annotations, resource_version
-    KubernetesWatchType = Tuple[str, str, Optional[TaskInstanceState], Dict[str, str], str]
+    KubernetesWatchType = Tuple[str, str, Optional[Union[TaskInstanceState, str]], Dict[str, str], str]
 
 ALL_NAMESPACES = "ALL_NAMESPACES"
 POD_EXECUTOR_DONE_KEY = "airflow_executor_done"

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/executors/kubernetes_executor_utils.py

@@ -40,6 +40,7 @@ from airflow.utils.state import TaskInstanceState
 
 try:
     from airflow.providers.cncf.kubernetes.executors.kubernetes_executor_types import (
+        ADOPTED,
         ALL_NAMESPACES,
         POD_EXECUTOR_DONE_KEY,
     )
@@ -220,7 +221,13 @@ class KubernetesJobWatcher(multiprocessing.Process, LoggingMixin):
         pod = event["object"]
         annotations_string = annotations_for_logging_task_metadata(annotations)
         """Process status response."""
-        if status == "Pending":
+        if event["type"] == "DELETED" and not pod.metadata.deletion_timestamp:
+            # This will happen only when the task pods are adopted by another executor.
+            # So, there is no change in the pod state.
+            # However, need to free the executor slot from the current executor.
+            self.log.info("Event: pod %s adopted, annotations: %s", pod_name, annotations_string)
+            self.watcher_queue.put((pod_name, namespace, ADOPTED, annotations, resource_version))
+        elif status == "Pending":
             # deletion_timestamp is set by kube server when a graceful deletion is requested.
             # since kube server have received request to delete pod set TI state failed
             if event["type"] == "DELETED" and pod.metadata.deletion_timestamp:

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/get_provider_info.py

@@ -28,8 +28,9 @@ def get_provider_info():
         "name": "Kubernetes",
         "description": "`Kubernetes <https://kubernetes.io/>`__\n",
         "suspended": False,
-        "source-date-epoch": 1701983367,
+        "source-date-epoch": 1703288121,
         "versions": [
+            "7.12.0",
             "7.11.0",
             "7.10.0",
             "7.9.0",

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/hooks/kubernetes.py

@@ -86,8 +86,8 @@ class KubernetesHook(BaseHook, PodOperatorHookProtocol):
 
     DEFAULT_NAMESPACE = "default"
 
-    @staticmethod
-    def get_connection_form_widgets() -> dict[str, Any]:
+    @classmethod
+    def get_connection_form_widgets(cls) -> dict[str, Any]:
         """Return connection widgets to add to connection form."""
         from flask_appbuilder.fieldwidgets import BS3TextFieldWidget
         from flask_babel import lazy_gettext
@@ -111,8 +111,8 @@ class KubernetesHook(BaseHook, PodOperatorHookProtocol):
             ),
         }
 
-    @staticmethod
-    def get_ui_field_behaviour() -> dict[str, Any]:
+    @classmethod
+    def get_ui_field_behaviour(cls) -> dict[str, Any]:
         """Return custom field behaviour."""
         return {
             "hidden_fields": ["host", "schema", "login", "password", "port", "extra"],

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/kubernetes_helper_functions.py

@@ -34,6 +34,8 @@ log = logging.getLogger(__name__)
 
 alphanum_lower = string.ascii_lowercase + string.digits
 
+POD_NAME_MAX_LENGTH = 63  # Matches Linux kernel's HOST_NAME_MAX default value minus 1.
+
 
 def rand_str(num):
     """Generate random lowercase alphanumeric string of length num.
@@ -43,7 +45,7 @@ def rand_str(num):
     return "".join(secrets.choice(alphanum_lower) for _ in range(num))
 
 
-def add_pod_suffix(*, pod_name: str, rand_len: int = 8, max_len: int = 80) -> str:
+def add_pod_suffix(*, pod_name: str, rand_len: int = 8, max_len: int = POD_NAME_MAX_LENGTH) -> str:
     """Add random string to pod name while staying under max length.
 
     :param pod_name: name of the pod
@@ -59,16 +61,12 @@ def create_pod_id(
     dag_id: str | None = None,
     task_id: str | None = None,
     *,
-    max_length: int = 80,
+    max_length: int = POD_NAME_MAX_LENGTH,
     unique: bool = True,
 ) -> str:
     """
     Generate unique pod ID given a dag_id and / or task_id.
 
-    The default of 80 for max length is somewhat arbitrary, mainly a balance between
-    content and not overwhelming terminal windows of reasonable width. The true
-    upper limit is 253, and this is enforced in construct_pod.
-
     :param dag_id: DAG ID
     :param task_id: Task ID
     :param max_length: max number of characters

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/operators/pod.py

@@ -51,6 +51,7 @@ from airflow.providers.cncf.kubernetes.backcompat.backwards_compat_converters im
     convert_volume_mount,
 )
 from airflow.providers.cncf.kubernetes.hooks.kubernetes import KubernetesHook
+from airflow.providers.cncf.kubernetes.kubernetes_helper_functions import POD_NAME_MAX_LENGTH
 from airflow.providers.cncf.kubernetes.pod_generator import PodGenerator
 from airflow.providers.cncf.kubernetes.triggers.pod import KubernetesPodTrigger
 from airflow.providers.cncf.kubernetes.utils import xcom_sidecar  # type: ignore[attr-defined]
@@ -92,7 +93,7 @@ def _rand_str(num):
     return "".join(secrets.choice(alphanum_lower) for _ in range(num))
 
 
-def _add_pod_suffix(*, pod_name, rand_len=8, max_len=253):
+def _add_pod_suffix(*, pod_name, rand_len=8, max_len=POD_NAME_MAX_LENGTH):
     """Add random string to pod name while staying under max len.
 
     TODO: when min airflow version >= 2.5, delete this function and import from kubernetes_helper_functions.
@@ -107,7 +108,7 @@ def _create_pod_id(
     dag_id: str | None = None,
     task_id: str | None = None,
     *,
-    max_length: int = 80,
+    max_length: int = POD_NAME_MAX_LENGTH,
     unique: bool = True,
 ) -> str:
     """
@@ -218,6 +219,7 @@ class KubernetesPodOperator(BaseOperator):
         /airflow/xcom/return.json in the container will also be pushed to an
         XCom when the container completes.
     :param pod_template_file: path to pod template file (templated)
+    :param pod_template_dict: pod template dictionary (templated)
     :param priority_class_name: priority class name for the launched Pod
     :param pod_runtime_info_envs: (Optional) A list of environment variables,
         to be set in the container.
@@ -267,6 +269,7 @@ class KubernetesPodOperator(BaseOperator):
         "labels",
         "config_file",
         "pod_template_file",
+        "pod_template_dict",
         "namespace",
         "container_resources",
         "volumes",
@@ -322,6 +325,7 @@ class KubernetesPodOperator(BaseOperator):
         log_events_on_failure: bool = False,
         do_xcom_push: bool = False,
         pod_template_file: str | None = None,
+        pod_template_dict: dict | None = None,
         priority_class_name: str | None = None,
         pod_runtime_info_envs: list[k8s.V1EnvVar] | None = None,
         termination_grace_period: int | None = None,
@@ -404,6 +408,7 @@ class KubernetesPodOperator(BaseOperator):
         self.log_events_on_failure = log_events_on_failure
         self.priority_class_name = priority_class_name
         self.pod_template_file = pod_template_file
+        self.pod_template_dict = pod_template_dict
         self.name = self._set_name(name)
         self.random_name_suffix = random_name_suffix
         self.termination_grace_period = termination_grace_period
@@ -672,6 +677,7 @@ class KubernetesPodOperator(BaseOperator):
         )
 
     def execute_complete(self, context: Context, event: dict, **kwargs):
+        self.log.debug("Triggered with event: %s", event)
         pod = None
         try:
             pod = self.hook.get_pod(
@@ -682,7 +688,11 @@ class KubernetesPodOperator(BaseOperator):
                 # fetch some logs when pod is failed
                 if self.get_logs:
                     self.write_logs(pod)
-                raise AirflowException(event["message"])
+                if "stack_trace" in event:
+                    message = f"{event['message']}\n{event['stack_trace']}"
+                else:
+                    message = event["message"]
+                raise AirflowException(message)
             elif event["status"] == "success":
                 # fetch some logs when pod is executed successfully
                 if self.get_logs:
@@ -892,6 +902,11 @@ class KubernetesPodOperator(BaseOperator):
             pod_template = pod_generator.PodGenerator.deserialize_model_file(self.pod_template_file)
             if self.full_pod_spec:
                 pod_template = PodGenerator.reconcile_pods(pod_template, self.full_pod_spec)
+        elif self.pod_template_dict:
+            self.log.debug("Pod template dict found, will parse for base pod")
+            pod_template = pod_generator.PodGenerator.deserialize_model_dict(self.pod_template_dict)
+            if self.full_pod_spec:
+                pod_template = PodGenerator.reconcile_pods(pod_template, self.full_pod_spec)
         elif self.full_pod_spec:
             pod_template = self.full_pod_spec
         else:
@@ -948,7 +963,7 @@ class KubernetesPodOperator(BaseOperator):
 
         if not pod.metadata.name:
             pod.metadata.name = _create_pod_id(
-                task_id=self.task_id, unique=self.random_name_suffix, max_length=80
+                task_id=self.task_id, unique=self.random_name_suffix, max_length=POD_NAME_MAX_LENGTH
             )
         elif self.random_name_suffix:
             # user has supplied pod name, we're just adding suffix

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/pod_generator.py

@@ -41,7 +41,11 @@ from airflow.exceptions import (
     AirflowException,
     RemovedInAirflow3Warning,
 )
-from airflow.providers.cncf.kubernetes.kubernetes_helper_functions import add_pod_suffix, rand_str
+from airflow.providers.cncf.kubernetes.kubernetes_helper_functions import (
+    POD_NAME_MAX_LENGTH,
+    add_pod_suffix,
+    rand_str,
+)
 from airflow.providers.cncf.kubernetes.pod_generator_deprecated import (
     PodDefaults,
     PodGenerator as PodGeneratorDeprecated,
@@ -380,11 +384,11 @@ class PodGenerator:
             - executor_config
             - dynamic arguments
         """
-        if len(pod_id) > 253:
+        if len(pod_id) > POD_NAME_MAX_LENGTH:
             warnings.warn(
-                "pod_id supplied is longer than 253 characters; truncating and adding unique suffix."
+                f"pod_id supplied is longer than {POD_NAME_MAX_LENGTH} characters; truncating and adding unique suffix."
             )
-            pod_id = add_pod_suffix(pod_name=pod_id, max_len=253)
+            pod_id = add_pod_suffix(pod_name=pod_id, max_len=POD_NAME_MAX_LENGTH)
         try:
             image = pod_override_object.spec.containers[0].image  # type: ignore
             if not image:

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/triggers/pod.py

@@ -18,9 +18,11 @@ from __future__ import annotations
 
 import asyncio
 import datetime
+import traceback
 import warnings
 from asyncio import CancelledError
 from enum import Enum
+from functools import cached_property
 from typing import TYPE_CHECKING, Any, AsyncIterator
 
 from airflow.exceptions import AirflowProviderDeprecationWarning
@@ -115,7 +117,6 @@ class KubernetesPodTrigger(BaseTrigger):
             self.on_finish_action = OnFinishAction(on_finish_action)
             self.should_delete_pod = self.on_finish_action == OnFinishAction.DELETE_POD
 
-        self._hook: AsyncKubernetesHook | None = None
         self._since_time = None
 
     def serialize(self) -> tuple[str, dict[str, Any]]:
@@ -141,11 +142,10 @@ class KubernetesPodTrigger(BaseTrigger):
 
     async def run(self) -> AsyncIterator[TriggerEvent]:  # type: ignore[override]
         """Get current pod status and yield a TriggerEvent."""
-        hook = self._get_async_hook()
         self.log.info("Checking pod %r in namespace %r.", self.pod_name, self.pod_namespace)
         try:
             while True:
-                pod = await hook.get_pod(
+                pod = await self.hook.get_pod(
                     name=self.pod_name,
                     namespace=self.pod_namespace,
                 )
@@ -205,13 +205,13 @@ class KubernetesPodTrigger(BaseTrigger):
             # That means that task was marked as failed
             if self.get_logs:
                 self.log.info("Outputting container logs...")
-                await self._get_async_hook().read_logs(
+                await self.hook.read_logs(
                     name=self.pod_name,
                     namespace=self.pod_namespace,
                 )
             if self.on_finish_action == OnFinishAction.DELETE_POD:
                 self.log.info("Deleting pod...")
-                await self._get_async_hook().delete_pod(
+                await self.hook.delete_pod(
                     name=self.pod_name,
                     namespace=self.pod_namespace,
                 )
@@ -231,18 +231,22 @@ class KubernetesPodTrigger(BaseTrigger):
                     "namespace": self.pod_namespace,
                     "status": "error",
                     "message": str(e),
+                    "stack_trace": traceback.format_exc(),
                 }
             )
 
     def _get_async_hook(self) -> AsyncKubernetesHook:
-        if self._hook is None:
-            self._hook = AsyncKubernetesHook(
-                conn_id=self.kubernetes_conn_id,
-                in_cluster=self.in_cluster,
-                config_file=self.config_file,
-                cluster_context=self.cluster_context,
-            )
-        return self._hook
+        # TODO: Remove this method when the min version of kubernetes provider is 7.12.0 in Google provider.
+        return AsyncKubernetesHook(
+            conn_id=self.kubernetes_conn_id,
+            in_cluster=self.in_cluster,
+            config_file=self.config_file,
+            cluster_context=self.cluster_context,
+        )
+
+    @cached_property
+    def hook(self) -> AsyncKubernetesHook:
+        return self._get_async_hook()
 
     def define_container_state(self, pod: V1Pod) -> ContainerState:
         pod_containers = pod.status.container_statuses

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/airflow/providers/cncf/kubernetes/utils/pod_manager.py

@@ -37,7 +37,7 @@ from kubernetes.stream import stream as kubernetes_stream
 from pendulum import DateTime
 from pendulum.parsing.exceptions import ParserError
 from typing_extensions import Literal
-from urllib3.exceptions import HTTPError as BaseHTTPError
+from urllib3.exceptions import HTTPError, TimeoutError
 
 from airflow.exceptions import AirflowException, AirflowProviderDeprecationWarning
 from airflow.providers.cncf.kubernetes.pod_generator import PodDefaults
@@ -395,7 +395,7 @@ class PodManager(LoggingMixin):
         :meta private:
         """
 
-        def consume_logs(*, since_time: DateTime | None = None) -> DateTime | None:
+        def consume_logs(*, since_time: DateTime | None = None) -> tuple[DateTime | None, Exception | None]:
             """
             Try to follow container logs until container completes.
 
@@ -404,7 +404,18 @@ class PodManager(LoggingMixin):
 
             Returns the last timestamp observed in logs.
             """
+            exception = None
             last_captured_timestamp = None
+            # We timeout connections after 30 minutes because otherwise they can get
+            # stuck forever. The 30 is somewhat arbitrary.
+            # As a consequence, a TimeoutError will be raised no more than 30 minutes
+            # after starting read.
+            connection_timeout = 60 * 30
+            # We set a shorter read timeout because that helps reduce *connection* timeouts
+            # (since the connection will be restarted periodically). And with read timeout,
+            # we don't need to worry about either duplicate messages or losing messages; we
+            # can safely resume from a few seconds later
+            read_timeout = 60 * 5
             try:
                 logs = self.read_pod_logs(
                     pod=pod,
@@ -415,6 +426,7 @@ class PodManager(LoggingMixin):
                     ),
                     follow=follow,
                     post_termination_timeout=post_termination_timeout,
+                    _request_timeout=(connection_timeout, read_timeout),
                 )
                 message_to_log = None
                 message_timestamp = None
@@ -447,29 +459,37 @@ class PodManager(LoggingMixin):
                             self._progress_callback(line)
                     self.log.info("[%s] %s", container_name, message_to_log)
                     last_captured_timestamp = message_timestamp
-            except BaseHTTPError:
+            except TimeoutError as e:
+                # in case of timeout, increment return time by 2 seconds to avoid
+                # duplicate log entries
+                if val := (last_captured_timestamp or since_time):
+                    return val.add(seconds=2), e
+            except HTTPError as e:
+                exception = e
                 self.log.exception(
                     "Reading of logs interrupted for container %r; will retry.",
                     container_name,
                 )
-            return last_captured_timestamp or since_time
+            return last_captured_timestamp or since_time, exception
 
         # note: `read_pod_logs` follows the logs, so we shouldn't necessarily *need* to
         # loop as we do here. But in a long-running process we might temporarily lose connectivity.
         # So the looping logic is there to let us resume following the logs.
         last_log_time = since_time
         while True:
-            last_log_time = consume_logs(since_time=last_log_time)
+            last_log_time, exc = consume_logs(since_time=last_log_time)
             if not self.container_is_running(pod, container_name=container_name):
                 return PodLoggingStatus(running=False, last_log_time=last_log_time)
             if not follow:
                 return PodLoggingStatus(running=True, last_log_time=last_log_time)
             else:
-                self.log.warning(
-                    "Pod %s log read interrupted but container %s still running",
-                    pod.metadata.name,
-                    container_name,
-                )
+                # a timeout is a normal thing and we ignore it and resume following logs
+                if not isinstance(exc, TimeoutError):
+                    self.log.warning(
+                        "Pod %s log read interrupted but container %s still running",
+                        pod.metadata.name,
+                        container_name,
+                    )
                 time.sleep(1)
 
     def _reconcile_requested_log_containers(
@@ -610,6 +630,7 @@ class PodManager(LoggingMixin):
         since_seconds: int | None = None,
         follow=True,
         post_termination_timeout: int = 120,
+        **kwargs,
     ) -> PodLogsConsumer:
         """Read log from the POD."""
         additional_kwargs = {}
@@ -618,6 +639,7 @@ class PodManager(LoggingMixin):
 
         if tail_lines:
             additional_kwargs["tail_lines"] = tail_lines
+        additional_kwargs.update(**kwargs)
 
         try:
             logs = self._client.read_namespaced_pod_log(
@@ -629,7 +651,7 @@ class PodManager(LoggingMixin):
                 _preload_content=False,
                 **additional_kwargs,
             )
-        except BaseHTTPError:
+        except HTTPError:
             self.log.exception("There was an error reading the kubernetes API.")
             raise
 
@@ -658,7 +680,7 @@ class PodManager(LoggingMixin):
             return self._client.list_namespaced_event(
                 namespace=pod.metadata.namespace, field_selector=f"involvedObject.name={pod.metadata.name}"
             )
-        except BaseHTTPError as e:
+        except HTTPError as e:
             raise AirflowException(f"There was an error reading the kubernetes API: {e}")
 
     @tenacity.retry(stop=tenacity.stop_after_attempt(3), wait=tenacity.wait_exponential(), reraise=True)
@@ -666,7 +688,7 @@ class PodManager(LoggingMixin):
         """Read POD information."""
         try:
             return self._client.read_namespaced_pod(pod.metadata.name, pod.metadata.namespace)
-        except BaseHTTPError as e:
+        except HTTPError as e:
             raise AirflowException(f"There was an error reading the kubernetes API: {e}")
 
     def await_xcom_sidecar_container_start(self, pod: V1Pod) -> None:

{apache_airflow_providers_cncf_kubernetes-7.11.0 → apache_airflow_providers_cncf_kubernetes-7.12.0}/pyproject.toml

@@ -28,7 +28,7 @@ build-backend = "flit_core.buildapi"
 
 [project]
 name = "apache-airflow-providers-cncf-kubernetes"
-version = "7.11.0"
+version = "7.12.0"
 description = "Provider package apache-airflow-providers-cncf-kubernetes for Apache Airflow"
 readme = "README.rst"
 authors = [
@@ -65,8 +65,8 @@ dependencies = [
 ]
 
 [project.urls]
-"Documentation" = "https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0"
-"Changelog" = "https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.11.0/changelog.html"
+"Documentation" = "https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0"
+"Changelog" = "https://airflow.apache.org/docs/apache-airflow-providers-cncf-kubernetes/7.12.0/changelog.html"
 "Bug Tracker" = "https://github.com/apache/airflow/issues"
 "Source Code" = "https://github.com/apache/airflow"
 "Slack Chat" = "https://s.apache.org/airflow-slack"