airflow.providers.cncf.kubernetes.triggers.pod

Classes

ContainerState

Possible container states.

KubernetesPodTrigger

KubernetesPodTrigger run on the trigger worker to check the state of Pod.

Module Contents

class airflow.providers.cncf.kubernetes.triggers.pod.ContainerState[source]

Bases: str, enum.Enum

Possible container states.

See https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#pod-phase.

WAITING = 'waiting'[source]
RUNNING = 'running'[source]
TERMINATED = 'terminated'[source]
FAILED = 'failed'[source]
UNDEFINED = 'undefined'[source]
class airflow.providers.cncf.kubernetes.triggers.pod.KubernetesPodTrigger(pod_name, pod_namespace, trigger_start_time, base_container_name, kubernetes_conn_id=None, connection_extras=None, poll_interval=2, cluster_context=None, config_dict=None, in_cluster=None, get_logs=True, startup_timeout=120, startup_check_interval=5, schedule_timeout=120, on_finish_action='delete_pod', on_kill_action='delete_pod', termination_grace_period=None, last_log_time=None, logging_interval=None, trigger_kwargs=None)[source]

Bases: airflow.triggers.base.BaseTrigger

KubernetesPodTrigger run on the trigger worker to check the state of Pod.

Parameters:

  • pod_name (str) – The name of the pod.

  • pod_namespace (str) – The namespace of the pod.

  • kubernetes_conn_id (str | None) – The kubernetes connection id for the Kubernetes cluster.

  • cluster_context (str | None) – Context that points to kubernetes cluster.

  • config_dict (dict | None) – Content of kubeconfig file in dict format.

  • poll_interval (float) – Polling period in seconds to check for the status.

  • trigger_start_time (datetime.datetime) – time in Datetime format when the trigger was started

  • in_cluster (bool | None) – run kubernetes client with in_cluster configuration.

  • get_logs (bool) – get the stdout of the container as logs of the tasks.

  • startup_timeout (int) – timeout in seconds to start up the pod.

  • startup_check_interval (float) – interval in seconds to check if the pod has already started.

  • schedule_timeout (int) – timeout in seconds to schedule pod in cluster.

  • on_finish_action (str) – What to do when the pod reaches its final state, or the execution is interrupted. If “delete_pod”, the pod will be deleted regardless its state; if “delete_succeeded_pod”, only succeeded pod will be deleted. You can set to “keep_pod” to keep the pod.

  • on_kill_action (str) – What to do when the trigger is cancelled (e.g. when a deferred task is manually marked as success/failed). If “delete_pod” (default), the pod will be deleted. If “keep_pod”, the pod will not be deleted.

  • termination_grace_period (int | None) – Optional grace period in seconds for pod termination during cleanup.

  • logging_interval (int | None) – number of seconds to wait before kicking it back to the operator to print latest logs. If None will wait until container done.

  • last_log_time (pendulum.DateTime | None) – where to resume logs from

  • trigger_kwargs (dict | None) – additional keyword parameters to send in the event

pod_name[source]
pod_namespace[source]
trigger_start_time[source]
base_container_name[source]
kubernetes_conn_id = None[source]
connection_extras = None[source]
poll_interval = 2[source]
cluster_context = None[source]
config_dict = None[source]
in_cluster = None[source]
get_logs = True[source]
startup_timeout = 120[source]
startup_check_interval = 5[source]
schedule_timeout = 120[source]
last_log_time = None[source]
logging_interval = None[source]
on_finish_action[source]
on_kill_action[source]
termination_grace_period = None[source]
trigger_kwargs[source]
serialize()[source]

Serialize KubernetesCreatePodTrigger arguments and classpath.

async run()[source]

Get current pod status and yield a TriggerEvent.

property hook: airflow.providers.cncf.kubernetes.hooks.kubernetes.AsyncKubernetesHook[source]
property pod_manager: airflow.providers.cncf.kubernetes.utils.pod_manager.AsyncPodManager[source]
get_task_instance(session)[source]

Get the task instance for this trigger from the database (Airflow 2.x only).

async get_task_state()[source]

Get the current state of the task instance.

async safe_to_cancel()[source]

Whether it is safe to delete the pod during trigger cleanup.

Used only on Airflow < 3.3.0 where the triggerer does not invoke on_kill() for user kills. Deletion is NOT safe when the task is still in DEFERRED state (triggerer restart).

async on_kill()[source]

Delete the pod when the trigger is cancelled by a user action.

The triggerer invokes this for user-initiated kills on Airflow 3.3.0+ only; on older versions use cleanup() and safe_to_cancel() instead.

async cleanup()[source]

Clean up the pod when the trigger exits.

On Airflow 3.3.0+ pod deletion on user kill is handled in on_kill() only; this avoids deleting pods on triggerer restart. On older Airflow versions, cleanup() still uses safe_to_cancel() because on_kill() is not wired for user kills.

define_container_state(pod)[source]
static should_wait(pod_phase, container_state)[source]

Was this entry helpful?