NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #1385 most downloaded on PyPI
Provider package apache-airflow-providers-cncf-kubernetes for Apache Airflow
Last release 4 days ago
14 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 60 of the last 60 stable releases
6 versions withdrawn
withdrawn after publishing
6 years old
235 releases · first in 2020
- Ignore mypy errors for deprecated executors
Release Date: 2025-07-06
Fix deferrable mode for SparkKubernetesOperator (#51956)
Reduce log noise from K8s Pod Operator for XCom (#51812)
Prevent legacy static hybrid executors to be running in Airflow 3 (#51760)
Move 'BaseHook' implementation to task SDK (#51873)
Update Kubernetes provider for Airflow 3.0 compatibility (#52664)
Upgrade ruff to latest version (0.12.1) (#52562)
Drop support for Python 3.9 (#52072)
Use BaseSensorOperator from task sdk in providers (#52296)
Move type-ignores up one line (#52195)
Ignore mypy errors for deprecated executors (#52187)
Fix deferrable mode for SparkKubernetesOperator (#51956)
Reduce log noise from K8s Pod Operator for XCom (#51812)
Prevent legacy static hybrid executors to be running in Airflow 3 (#51760)
Move 'BaseHook' implementation to task SDK (#51873)
Update Kubernetes provider for Airflow 3.0 compatibility (#52664)
Upgrade ruff to latest version (0.12.1) (#52562)
Drop support for Python 3.9 (#52072)
Use BaseSensorOperator from task sdk in providers (#52296)
Move type-ignores up one line (#52195)
Ignore mypy errors for deprecated executors (#52187)
One column per quarter.
Nothing published for this version
- KubernetesPodOperator push xcom after failed pod
Release Date: 2025-06-18
KubernetesPodOperator push xcom after failed pod (#51475)
[KubernetesPodOperator] Reads Kubernetes events and writes them into log (#50192)
Added detection of terminated pod during get_or_create_pod (#51158)
Add xcom sidecar terminated detection (#51144)
Fix 'EksPodOperator' in deferrable mode (#51255)
Fix a bug where Kube config "worker_pod_pending_fatal_container_state_reasons" is parsed wrongly (#50931)
Relax Kubernetes client version support to <33.0.0 (#51195)
Adjust code path in executors to be compatible with airflow task run (#51009)
Port ''ti.run'' to Task SDK execution path (#50141)
Bump some provider dependencies for faster resolution (#51727)
Prevent legacy static hybrid executors to be running in Airflow 3 (#51733)
Update the executor and provider doc to highlight the two statically coded hybrid executors are no longer supported in Airflow 3.0.0+ (#51715)
Update operators.rst - fix typo Kubernetes Pod Operator doc (#51574)
KubernetesPodOperator push xcom after failed pod (#51475)
[KubernetesPodOperator] Reads Kubernetes events and writes them into log (#50192)
Added detection of terminated pod during get_or_create_pod (#51158)
Add xcom sidecar terminated detection (#51144)
Fix 'EksPodOperator' in deferrable mode (#51255)
Fix a bug where Kube config "worker_pod_pending_fatal_container_state_reasons" is parsed wrongly (#50931)
Relax Kubernetes client version support to <33.0.0 (#51195)
Adjust code path in executors to be compatible with airflow task run (#51009)
Port ''ti.run'' to Task SDK execution path (#50141)
Bump some provider dependencies for faster resolution (#51727)
Prevent legacy static hybrid executors to be running in Airflow 3 (#51733)
Update the executor and provider doc to highlight the two statically coded hybrid executors are no longer supported in Airflow 3.0.0+ (#51715)
Update operators.rst - fix typo Kubernetes Pod Operator doc (#51574)
Nothing published for this version
This release of provider is only available for Airflow 2.10+ as explained in the Apache Airflow providers support policy < https://github.com/apache/a
Release Date: 2025-05-21
Note
This release of provider is only available for Airflow 2.10+ as explained in the Apache Airflow providers support policy < https://github.com/apache/airflow/blob/main/PROVIDERS.rst#minimum-supported-version-of-airflow-for-community-managed-providers >_.
Add fast fail for ErrImagePull and InvalidImageName for KubernetesPodOperator (#49867)
KubernetesPodOperator uses different timeouts to check for schedule timeout and startup timeout (#49784)
Add configurable automountServiceAccountToken for the KubernetesPodOperator (#50223)
Add 'test_connection' method to 'KubernetesHook' (#47881)
Add '@task.kuberenetes_cmd' (#46913)
Pass workload to KubernetesExecutor using command args instead of init container (#50448)
Fix KubernetesPodOperator AutomountServiceAccountToken from Pod Template (#50800)
Fix broken imports of cncf.kubernetes for Airflow 2 (#50651)
Move SQS message queue to Amazon provider (#50057)
Remove AIRFLOW_2_10_PLUS conditions (#49877)
Bump min Airflow version in providers to 2.10 (#49843)
Note
This release of provider is only available for Airflow 2.10+ as explained in the Apache Airflow providers support policy <https://github.com/apache/airflow/blob/main/PROVIDERS.rst#minimum-supported-version-of-airflow-for-community-managed-providers>_.
Add fast fail for ErrImagePull and InvalidImageName for KubernetesPodOperator (#49867)
KubernetesPodOperator uses different timeouts to check for schedule timeout and startup timeout (#49784)
Add configurable automountServiceAccountToken for the KubernetesPodOperator (#50223)
Add 'test_connection' method to 'KubernetesHook' (#47881)
Add '@task.kuberenetes_cmd' (#46913)
Pass workload to KubernetesExecutor using command args instead of init container (#50448)
Fix KubernetesPodOperator AutomountServiceAccountToken from Pod Template (#50800)
Fix broken imports of cncf.kubernetes for Airflow 2 (#50651)
Move SQS message queue to Amazon provider (#50057)
Remove AIRFLOW_2_10_PLUS conditions (#49877)
Bump min Airflow version in providers to 2.10 (#49843)
Nothing published for this version
Nothing published for this version
- Remove 'subdir' arg from CLI commands
Release Date: 2025-04-19
Remove 'subdir' arg from CLI commands (#49317)
Use contextlib.suppress(exception) instead of try-except-pass and add SIM105 ruff rule (#49251)
remove superfluous else block (#49199)
Remove unused db method in k8s provider (#49186)
Nothing published for this version
- Make '@task' import from airflow.sdk
Release Date: 2025-04-14
Make '@task' import from airflow.sdk (#48896)
Update ECS executor to support Task SDK (#48513)
Make '@task' import from airflow.sdk (#48896)
Update ECS executor to support Task SDK (#48513)
Nothing published for this version
- Handle missing container_statuses in pod_manager.get_container_status
Release Date: 2025-04-09
Handle missing container_statuses in pod_manager.get_container_status (#47936)
Remove change_sensor_mode_to_reschedule from base executor (#48649)
No need to exclude executor_config specifically for KE (#48826)
Handle missing container_statuses in pod_manager.get_container_status (#47936)
Remove change_sensor_mode_to_reschedule from base executor (#48649)
No need to exclude executor_config specifically for KE (#48826)
Nothing published for this version
- KubernetesPodOperator: add base_container_name to the templated fields
Release Date: 2025-03-31
KubernetesPodOperator: add base_container_name to the templated fields (#47864)
cncf.kubernetes: make the base container status check polling interval configurable (#47774)
Switch to non-interactive mode where stdin is disabled (#47568)
Retry k8s API requests in 'KubernetesPodTrigger' (#47187)
Remove extra method for async hook getting (#47313)
KubernetesPodOperator: add base_container_name to the templated fields (#47864)
cncf.kubernetes: make the base container status check polling interval configurable (#47774)
Switch to non-interactive mode where stdin is disabled (#47568)
Retry k8s API requests in 'KubernetesPodTrigger' (#47187)
Remove extra method for async hook getting (#47313)
Nothing published for this version
Nothing published for this version
- Get rid of google-re2 as dependency
Release Date: 2025-03-13
Get rid of google-re2 as dependency (#47493)
AIP-72: Port over executor_config for K8sExecutor when using task sdk (#46926)
Update templated fields in KPO doc-string + wording (#46970)
Upgrade flit to 3.11.0 (#46938)
Get rid of google-re2 as dependency (#47493)
AIP-72: Port over executor_config for K8sExecutor when using task sdk (#46926)
Update templated fields in KPO doc-string + wording (#46970)
Upgrade flit to 3.11.0 (#46938)
Nothing published for this version
This version contains no code changes. It was released to replace a previous version that was yanked due to a packaging issue.
Release Date: 2025-02-26
Note
This version contains no code changes. It was released to replace a previous version that was yanked due to a packaging issue.
Nothing published for this version
- Remove overlooked 'pod_generator_deprecated.py'
Release Date: 2025-02-21
KubernetesPodOperator new callbacks and allow multiple callbacks (#44357)
(fix): Render a pod spec using the pod_template_file override, if passed to the executor (#46374)
fix: kpo async kube_config_path (#45571)
Allow passing empty labels in the spark kubernetes driver config (#45864)
fix: spark operator label (#45353)
Allow internal retries when pending k8s pod is deleted (#45184)
fix: correctly shutdown xcom after job is complete (#45455)
Fix kubernetes executor watcher kube_client_request_args shadowing (#45528)
Fix wrong operator name in doc string of @task.kubernetes (#46873)
AIP-83 amendment: Add logic for generating run_id when logical date is None. (#46616)
Remove overlooked 'pod_generator_deprecated.py' (#46587)
More meaningful '@task.kubernetes' pod naming (#46535)
chore(kubernetes): reduce redundant code (#46574)
Add dynamic task mapping into TaskSDK runtime (#46032)
Expand KPO template_fields, fix Spark k8s operator tests (#46268)
Swap CeleryExecutor over to use TaskSDK for execution. (#46265)
Move Pod*Exceptions to separate module (#45759)
Add other instances of Context type hints (#45657)
AIP-72: Support better type-hinting for Context dict in SDK (#45583)
AIP-72: Swap KubernetesExecutor to use taskSDK for execution (#46860)
KubernetesPodOperator new callbacks and allow multiple callbacks (#44357)
(fix): Render a pod spec using the pod_template_file override, if passed to the executor (#46374)
fix: kpo async kube_config_path (#45571)
Allow passing empty labels in the spark kubernetes driver config (#45864)
fix: spark operator label (#45353)
Allow internal retries when pending k8s pod is deleted (#45184)
fix: correctly shutdown xcom after job is complete (#45455)
Fix kubernetes executor watcher kube_client_request_args shadowing (#45528)
Fix wrong operator name in doc string of @task.kubernetes (#46873)
AIP-83 amendment: Add logic for generating run_id when logical date is None. (#46616)
Remove overlooked 'pod_generator_deprecated.py' (#46587)
More meaningful '@task.kubernetes' pod naming (#46535)
chore(kubernetes): reduce redundant code (#46574)
Add dynamic task mapping into TaskSDK runtime (#46032)
Expand KPO template_fields, fix Spark k8s operator tests (#46268)
Swap CeleryExecutor over to use TaskSDK for execution. (#46265)
Move Pod*Exceptions to separate module (#45759)
Add other instances of Context type hints (#45657)
AIP-72: Support better type-hinting for Context dict in SDK (#45583)
AIP-72: Swap KubernetesExecutor to use taskSDK for execution (#46860)
This release of provider is only available for Airflow 2.9+ as explained in the Apache Airflow providers support policy .
Release Date: 2024-12-26
Note
This release of provider is only available for Airflow 2.9+ as explained in the Apache Airflow providers support policy .
Add logging support for init containers in KubernetesPodOperator (#42498) (#43853)
Implement KubernetesInstallKueueOperator + (#44568)
support grouping of log lines for KubernetesPodOperator (#44428)
Compare k8s executor against alias, not full ExecutorName repr (#44967)
Fix failing KubernetesExecutor (#44931)
Remove references to AIRFLOW_V_2_9_PLUS (#44987)
Bump minimum Airflow version in providers to Airflow 2.9.0 (#44956)
Consistent way of checking Airflow version in providers (#44686)
Fix yoda-conditions (#44466)
Note
This release of provider is only available for Airflow 2.9+ as explained in the Apache Airflow providers support policy.
Add logging support for init containers in KubernetesPodOperator (#42498) (#43853)
Implement KubernetesInstallKueueOperator + (#44568)
support grouping of log lines for KubernetesPodOperator (#44428)
Compare k8s executor against alias, not full ExecutorName repr (#44967)
Fix failing KubernetesExecutor (#44931)
Remove references to AIRFLOW_V_2_9_PLUS (#44987)
Bump minimum Airflow version in providers to Airflow 2.9.0 (#44956)
Consistent way of checking Airflow version in providers (#44686)
Fix yoda-conditions (#44466)
Nothing published for this version
Nothing published for this version
- Bugfix KubernetesJobOperator.on_kill() by removing unexpected 'job' argument from the k8s job client method 'delete_namespaced_job' invocation
Release Date: 2024-11-27
Bugfix KubernetesJobOperator.on_kill() by removing unexpected 'job' argument from the k8s job client method 'delete_namespaced_job' invocation (#44131)
Re-queue tassk when they are stuck in queued (#43520)
Rename execution_date to logical_date across codebase (#43902)
Bugfix KubernetesJobOperator.on_kill() by removing unexpected 'job' argument from the k8s job client method 'delete_namespaced_job' invocation (#44131)
Re-queue tassk when they are stuck in queued (#43520)
Rename execution_date to logical_date across codebase (#43902)
Nothing published for this version
All deprecated classes, parameters and features have been removed from the Kubernetes provider package. The following breaking changes were introduced…
Release Date: 2024-11-18
Warning
All deprecated classes, parameters and features have been removed from the Kubernetes provider package. The following breaking changes were introduced:
Remove add_pod_suffix method from kubernetes_helper_functions.py . Use add_unique_suffix instead.
Remove make_unique_pod_id method from PodGenerator . Use add_unique_suffix in kubernetes_helper_functions instead.
Remove create_pod_id method from kubernetes_helper_functions.py . Use create_unique_id instead.
Remove gen_pod method from PodGenerator .
Remove add_xcom_sidecar method from PodGenerator . Use airflow.providers.cncf.kubernetes.utils.xcom_sidecar.add_xcom_sidecar instead.
Remove the option to using a dictionary for the executor_config from_obj function in PodGenerator . Use a kubernetes.client.models.V1Pod class with a “pod_override” key.
Remove from_legacy_obj method from PodGenerator .
Remove airflow.providers.cncf.kubernetes.pod_launcher_deprecated module. Use airflow.providers.cncf.kubernetes.utils.pod_manager instead.
Remove airflow.providers.cncf.kubernetes.operators.kubernetes_pod . Use airflow.providers.cncf.kubernetes.operators.pod instead.
Remove is_delete_operator_pod parameters from KubernetesPodOperator . Use on_finish_action instead.
Remove progress_callback parameters from KubernetesPodOperator . Use callbacks instead.
Remove execute_complete method from KubernetesPodOperator . Use trigger_reentry instead.
Remove xcom_push parameter from SparkKubernetesOperator . Use do_xcom_push .
Remove should_delete_pod parameter from KubernetesPodTrigger . Use on_finish_action instead.
Remove progress_callback parameter from PodManager .
Remove follow_container_logs method from PodManager . Use fetch_container_logs instead.
Warning
Set the default value of namespace in @task.kubernetes to None , so it uses the cluster namespace when in_cluster is True. Be sure to specify a namespace when using this decorator. To retain the previous behavior, set namespace="default"
Remove deprecations cncf.kubernetes (#43689)
Change default value of 'namespace' in 'task.kubernetes' to be None (#43402)
Add random_name_suffix to SparkKubernetesOperator (#43847)
terminate kubernetes watch in case of unknown error (#43645)
Update 'namespace' priority for 'find_pod' function (#43762)
AIP-72: Remove DAG pickling (#43667)
Nothing published for this version
- (fix): HybridExecutor tasks of other executor rescheduled in kubernetes executor
Release Date: 2024-10-31
(fix): HybridExecutor tasks of other executor rescheduled in kubernetes executor (#43003)
(fix): use instance base_container_name to fetch logs on trigger_reentry (#42960)
Add kubernetes_conn_id to templated fields (#42786)
Allow node_selector templating in KPO (#43051)
Bump max version of Kubernetes Python SDK to 31.0.0 (#42907)
(fix): HybridExecutor tasks of other executor rescheduled in kubernetes executor (#43003)
(fix): use instance base_container_name to fetch logs on trigger_reentry (#42960)
Add kubernetes_conn_id to templated fields (#42786)
Allow node_selector templating in KPO (#43051)
Bump max version of Kubernetes Python SDK to 31.0.0 (#42907)
Nothing published for this version
Support for identifying pods by execution_date during the upgrade from Airflow 1 to 2 has been removed. This may result in duplicate pods being launch
Release Date: 2024-10-14
Warning
Support for identifying pods by execution_date during the upgrade from Airflow 1 to 2 has been removed. This may result in duplicate pods being launched for tasks originally started by Airflow 1, but only one of the task pods will succeed.
kubernetes executor cleanup_stuck_queued_tasks optimization (#41220)
All executors should inherit from BaseExecutor (#41904)
Fix mark as success when pod fails while fetching log (#42815)
Fix SparkKubernetesOperator spark name. (#42427)
KubernetesPodOperator never stops if credentials are refreshed (#42361)
Restructured 'await_xcom_sidecar_container_start' method. (#42504)
KubernetesHook kube_config extra can take dict (#41413)
Drop python3.8 support core and providers (#42766)
Remove airflow_version from k8s executor pod selector (#42751)
Nothing published for this version
- Deprecated configuration removed
Release Date: 2024-09-24
Deprecated configuration removed (#42129)
docstring update in AsyncKubernetesHook (#41929)
Remove Airflow 2.7 support from Kubernetes Provider (#41746)
remove deprecated soft_fail from providers (#41710)
Deprecated configuration removed (#42129)
docstring update in AsyncKubernetesHook (#41929)
Remove Airflow 2.7 support from Kubernetes Provider (#41746)
remove deprecated soft_fail from providers (#41710)
Nothing published for this version
- fix: 'KubernetesExecutor' failing the task in case the watcher receives an event with the reason ProviderFailed
Release Date: 2024-08-28
fix: 'KubernetesExecutor' failing the task in case the watcher receives an event with the reason ProviderFailed (#41186)
fix: 'do_xcom_push' and 'get_logs' functionality for KubernetesJobOperator (#40814)
fix: 'KubernetesHook' loading config file with '_is_in_cluster' set as False (#41464)
fix: Missing 'slots_occupied' in 'CeleryKubernetesExecutor' and 'LocalKubernetesExecutor' (#41602)
Nothing published for this version
This release of provider is only available for Airflow 2.8+ as explained in the Apache Airflow providers support policy .
Release Date: 2024-08-22
Note
This release of provider is only available for Airflow 2.8+ as explained in the Apache Airflow providers support policy .
Bump minimum Airflow version in providers to Airflow 2.8.0 (#41396)
Describe behaviour in docstring correctly (#41458)
Remove deprecated SubDags (#41390)
reorder docstring of 'SparkKubernetesOperator' (#41372)
Note
This release of provider is only available for Airflow 2.8+ as explained in the Apache Airflow providers support policy.
Bump minimum Airflow version in providers to Airflow 2.8.0 (#41396)
Describe behaviour in docstring correctly (#41458)
Remove deprecated SubDags (#41390)
reorder docstring of 'SparkKubernetesOperator' (#41372)
Nothing published for this version
- Pass content of kube/config file to triggerer as a dictionary
Release Date: 2024-08-06
Pass content of kube/config file to triggerer as a dictionary (#41178)
Fix confusing log message in kubernetes executor (#41035)
Fix ApiException handling when adopting completed pods (#41109)
Pass content of kube/config file to triggerer as a dictionary (#41178)
Fix confusing log message in kubernetes executor (#41035)
Fix ApiException handling when adopting completed pods (#41109)
Nothing published for this version
- Solve failing KPO task with task decorator and imported typing elements
Release Date: 2024-07-12
Solve failing KPO task with task decorator and imported typing elements (#40642)
Solve failing KPO task with task decorator and imported typing elements (#40642)
Nothing published for this version
- Fix Scheduler restarting due to too many completed pods in cluster
Release Date: 2024-06-27
Fix Scheduler restarting due to too many completed pods in cluster (#40183)
Bump minimum kubernetes lib version to kubernetes 29.0.0 (#40253)
Nothing published for this version
- Fixes KubernetesPodTrigger failing running pods with timeout
Release Date: 2024-06-09
Fixes KubernetesPodTrigger failing running pods with timeout (#40019)
Refresh properties on KubernetesPodOperator on token expiration also when logging (#39789)
Fix reattach_on_restart parameter for the sync mode (#39329)
Avoid resetting adopted task instances when retrying for kubernetes executor (#39406)
Include fatal reason for pod pending events (#39924)
Fixes KubernetesPodTrigger failing running pods with timeout (#40019)
Refresh properties on KubernetesPodOperator on token expiration also when logging (#39789)
Fix reattach_on_restart parameter for the sync mode (#39329)
Avoid resetting adopted task instances when retrying for kubernetes executor (#39406)
Include fatal reason for pod pending events (#39924)
Nothing published for this version
- Fix deprecated calls in 'cncf.kubernetes' provider
Release Date: 2024-06-03
Add timeout when watching pod events in k8s executor (#39551)
Add retry logic for KubernetesCreateResourceOperator and KubernetesJobOperator (#39201)
Fix deprecated calls in 'cncf.kubernetes' provider (#39381)
Handling exception getting logs when pods finish success (#39296)
fix wrong arguments in read_namespaced_pod_log call (#39874)
Move Kubernetes cli to provider package (#39587)
Remove compat code for 2.7.0 - its now the min Airflow version (#39591)
Simplify 'airflow_version' imports (#39497)
Replace pod_manager.read_pod_logs with client.read_namespaced_pod_log in KubernetesPodOperator._write_logs (#39112)
Add a warning message to KPO to warn of one second interval logs duplication (#39861)
Add timeout when watching pod events in k8s executor (#39551)
Add retry logic for KubernetesCreateResourceOperator and KubernetesJobOperator (#39201)
Fix deprecated calls in 'cncf.kubernetes' provider (#39381)
Handling exception getting logs when pods finish success (#39296)
fix wrong arguments in read_namespaced_pod_log call (#39874)
Move Kubernetes cli to provider package (#39587)
Remove compat code for 2.7.0 - its now the min Airflow version (#39591)
Simplify 'airflow_version' imports (#39497)
Replace pod_manager.read_pod_logs with client.read_namespaced_pod_log in KubernetesPodOperator._write_logs (#39112)
Add a warning message to KPO to warn of one second interval logs duplication (#39861)
Nothing published for this version
Nothing published for this version
- Moves airflow import in deprecated pod_generator to local
Release Date: 2024-05-06
Note
This release of provider is only available for Airflow 2.7+ as explained in the Apache Airflow providers support policy .
Add missing informative logs in KPO trigger until container has not finished (#37546)
fixes templated env vars for k8s pod operator (#39139)
Fix SparkKubernetesOperator when using initContainers (#38119)
Refresh properties on KubernetesPodOperator when k8s fails due to token expiration (#39325)
Bump minimum Airflow version in providers to Airflow 2.7.0 (#39240)
Remove unnecessary validation from cncf provider. (#39238)
Moves airflow import in deprecated pod_generator to local (#39062)
KPO xcom sidecar PodDefault usage (#38951)
Nothing published for this version
Release Date: 2024-04-21 ### Bug Fixes - Avoid logging empty line KPO
Release Date: 2024-04-21
Avoid logging empty line KPO (#38247)
Nothing published for this version
- removed usage of deprecated function for naming the pod in provider k8s pod.py
Release Date: 2024-04-13
KPO Add follow log in termination step (#38081)
Add GKECreateCustomResourceOperator and GKEDeleteCustomResourceOperator operators (#37616)
Implement deferrable mode for KubernetesJobOperator (#38251)
Create KubernetesPatchJobOperator operator (#38146)
Implement delete_on_status parameter for KubernetesDeleteJobOperator (#38458)
Implement deferrable mode for GKEStartJobOperator (#38454)
Use startup_check_interval_seconds instead of poll_interval to check pod while startup (#38075)
Implement wait_until_job_complete parameter for KubernetesJobOperator (#37998)
Use SIGINT signal number instead of signal name (#37905)
Fix spark operator log retrieval from driver (#38106)
Fix dynamic allocation specs handling for custom launcher (#38223)
Fix case if 'SparkKubernetesOperator.application_file' is templated file (#38035)
fix: reduce irrelevant error logs for pod events. (#37944)
Add GKEListJobsOperator and GKEDescribeJobOperator (#37598)
removed usage of deprecated function for naming the pod in provider k8s pod.py (#38638)
Create DeleteKubernetesJobOperator and GKEDeleteJobOperator operators (#37793)
Refactor GKE hooks (#38404)
KPO Add follow log in termination step (#38081)
Add GKECreateCustomResourceOperator and GKEDeleteCustomResourceOperator operators (#37616)
Implement deferrable mode for KubernetesJobOperator (#38251)
Create KubernetesPatchJobOperator operator (#38146)
Implement delete_on_status parameter for KubernetesDeleteJobOperator (#38458)
Implement deferrable mode for GKEStartJobOperator (#38454)
Use startup_check_interval_seconds instead of poll_interval to check pod while startup (#38075)
Implement wait_until_job_complete parameter for KubernetesJobOperator (#37998)
Use SIGINT signal number instead of signal name (#37905)
Fix spark operator log retrieval from driver (#38106)
Fix dynamic allocation specs handling for custom launcher (#38223)
Fix case if 'SparkKubernetesOperator.application_file' is templated file (#38035)
fix: reduce irrelevant error logs for pod events. (#37944)
Add GKEListJobsOperator and GKEDescribeJobOperator (#37598)
removed usage of deprecated function for naming the pod in provider k8s pod.py (#38638)
Create DeleteKubernetesJobOperator and GKEDeleteJobOperator operators (#37793)
Refactor GKE hooks (#38404)
Nothing published for this version
- Immediately fail the task in case of worker pod having a fatal container state
Release Date: 2024-03-08
Immediately fail the task in case of worker pod having a fatal container state (#37670)
Skip pod cleanup in case of pod creation failed (#37671)
Avoid non-recommended usage of logging (#37792)
Migrate executor docs to respective providers (#37728)
Immediately fail the task in case of worker pod having a fatal container state (#37670)
Skip pod cleanup in case of pod creation failed (#37671)
Avoid non-recommended usage of logging (#37792)
Migrate executor docs to respective providers (#37728)
Nothing published for this version
- feat: Switch all class, functions, methods deprecations to decorators
Release Date: 2024-02-20
In the case of Kube API exceeded quota errors, we have introduced the task_publish_max_retries flag to control the re-queuing task behavior. Changed the default behavior from unlimited retries to 0. The default behavior is no retries ( task_publish_max_retries==0 ). For unlimited retries, set task_publish_max_retries=-1 . For a fixed number of retries, set task_publish_max_retries to any positive integer.
Fix: The task is stuck in a queued state forever in case of pod launch errors (#36882)
Add logging_interval in KubernetesPodOperator to log container log periodically (#37279)
Create GKEStartJobOperator and KubernetesJobOperator (#36847)
Fix occasional attr-undefined for the python_kubernetes_script (#37318)
Fix hanging KPO on deferrable task with do_xcom_push (#37300)
Fix rendering 'SparkKubernetesOperator.template_body' (#37271)
Fix assignment of template field in 'init' in 'KubernetesPodOperator' (#37010)
KPO Maintain backward compatibility for execute_complete and trigger run method (#37454)
Fix KPO task hanging when pod fails to start within specified timeout (#37514)
Fix KeyError when KPO exits too soon (#37508)
feat: Switch all class, functions, methods deprecations to decorators (#36876)
Kubernetes version bump (#37040)
Add GKEStartKueueInsideClusterOperator (#37072)
Convert Kubernetes ApiException status code to string to ensure it's correctly checked (#37405)
In the case of Kube API exceeded quota errors, we have introduced the task_publish_max_retries flag to control the re-queuing task behavior. Changed the default behavior from unlimited retries to 0. The default behavior is no retries (task_publish_max_retries==0). For unlimited retries, set task_publish_max_retries=-1. For a fixed number of retries, set task_publish_max_retries to any positive integer.
Fix: The task is stuck in a queued state forever in case of pod launch errors (#36882)
Add logging_interval in KubernetesPodOperator to log container log periodically (#37279)
Create GKEStartJobOperator and KubernetesJobOperator (#36847)
Fix occasional attr-undefined for the python_kubernetes_script (#37318)
Fix hanging KPO on deferrable task with do_xcom_push (#37300)
Fix rendering 'SparkKubernetesOperator.template_body' (#37271)
Fix assignment of template field in '__init__' in 'KubernetesPodOperator' (#37010)
KPO Maintain backward compatibility for execute_complete and trigger run method (#37454)
Fix KPO task hanging when pod fails to start within specified timeout (#37514)
Fix KeyError when KPO exits too soon (#37508)
feat: Switch all class, functions, methods deprecations to decorators (#36876)
Kubernetes version bump (#37040)
Add GKEStartKueueInsideClusterOperator (#37072)
Convert Kubernetes ApiException status code to string to ensure it's correctly checked (#37405)
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →