En esta página, se describe cómo usar KubernetesPodOperator para implementar pods de Kubernetes desde Cloud Composer en el clúster de Google Kubernetes Engine que forma parte de tu entorno de Cloud Composer y garantizar que este tenga los recursos adecuados.
KubernetesPodOperator inicia Pods de Kubernetesen el clúster de tu entorno. En comparación, los operadores de Google Kubernetes Engine ejecutan pods de Kubernetes en un clúster específico, que puede ser un clúster independiente que no está relacionado con tu entorno. También puedes crear y borrar clústeres con los operadores de Google Kubernetes Engine.
KubernetesPodOperator es una buena opción si necesitas lo siguiente:
Dependencias de Python personalizadas que no están disponibles a través del repositorio público de PyPI.
Dependencias binarias que no están disponibles en la imagen de archivo del trabajador de Cloud Composer.
En esta página, se explica un ejemplo de DAG de Airflow que incluye las siguientes configuraciones de KubernetesPodOperator:
Recomendamos usar la versión más reciente de Cloud Composer.
Como mínimo, esta versión debe ser compatible como parte de la política de baja y asistencia.
Asegúrate de que tu entorno tenga recursos suficientes.
Iniciar pods en un entorno con escasez de recursos puede causar errores en el trabajador y el programador de Airflow.
Configura los recursos del entorno de Cloud Composer
Cuando creas un entorno de Cloud Composer, debes especificar sus
parámetros de rendimiento, incluidos los parámetros de rendimiento del clúster del
entorno. Iniciar pods de Kubernetes en el clúster de entorno puede causar competencia por los recursos del clúster, como CPU o memoria. Debido a que el programador y los trabajadores de Airflow están en el mismo clúster de GKE, los programadores y los trabajadores no funcionarán de forma correcta si la competencia provoca escasez de recursos.
Para evitar escasez de recursos, realiza una o más de las siguientes acciones:
Si aumentas la cantidad de nodos de tu entorno de Cloud Composer, aumentará la capacidad de procesamiento disponible para tus trabajadores. Este aumento no proporciona recursos adicionales para las tareas que requieren más CPU o RAM de las que proporciona el tipo de máquina especificado.
Durante la creación de entornos de Cloud Composer, puedes especificar un tipo de máquina. A fin de garantizar que haya recursos disponibles, especifica el tipo de máquina para el tipo de procesamiento que se produce en tu entorno de Cloud Composer.
Las siguientes secciones explican cada configuración de KubernetesPodOperator en el ejemplo. Para obtener información sobre cada variable de configuración, consulta la referencia de Airflow.
import datetime
from airflow import models
from airflow.kubernetes.secret import Secret
from airflow.providers.cncf.kubernetes.operators.kubernetes_pod import (
KubernetesPodOperator,
)
from kubernetes.client import models as k8s_models
# A Secret is an object that contains a small amount of sensitive data such as
# a password, a token, or a key. Such information might otherwise be put in a
# Pod specification or in an image; putting it in a Secret object allows for
# more control over how it is used, and reduces the risk of accidental
# exposure.
secret_env = Secret(
# Expose the secret as environment variable.
deploy_type="env",
# The name of the environment variable, since deploy_type is `env` rather
# than `volume`.
deploy_target="SQL_CONN",
# Name of the Kubernetes Secret
secret="airflow-secrets",
# Key of a secret stored in this Secret object
key="sql_alchemy_conn",
)
secret_volume = Secret(
deploy_type="volume",
# Path where we mount the secret as volume
deploy_target="/var/secrets/google",
# Name of Kubernetes Secret
secret="service-account",
# Key in the form of service account file name
key="service-account.json",
)
# If you are running Airflow in more than one time zone
# see https://airflow.apache.org/docs/apache-airflow/stable/timezone.html
# for best practices
YESTERDAY = datetime.datetime.now() - datetime.timedelta(days=1)
# If a Pod fails to launch, or has an error occur in the container, Airflow
# will show the task as failed, as well as contain all of the task logs
# required to debug.
with models.DAG(
dag_id="composer_sample_kubernetes_pod",
schedule_interval=datetime.timedelta(days=1),
start_date=YESTERDAY,
) as dag:
# Only name, namespace, image, and task_id are required to create a
# KubernetesPodOperator. In Cloud Composer, currently the operator defaults
# to using the config file found at `/home/airflow/composer_kube_config if
# no `config_file` parameter is specified. By default it will contain the
# credentials for Cloud Composer's Google Kubernetes Engine cluster that is
# created upon environment creation.
kubernetes_min_pod = KubernetesPodOperator(
# The ID specified for the task.
task_id="pod-ex-minimum",
# Name of task you want to run, used to generate Pod ID.
name="pod-ex-minimum",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# The namespace to run within Kubernetes, default namespace is
# `default`. In Composer 1 there is the potential for
# the resource starvation of Airflow workers and scheduler
# within the Cloud Composer environment,
# the recommended solution is to increase the amount of nodes in order
# to satisfy the computing requirements. Alternatively, launching pods
# into a custom namespace will stop fighting over resources,
# and using Composer 2 will mean the environment will autoscale.
namespace="default",
# Docker image specified. Defaults to hub.docker.com, but any fully
# qualified URLs will point to a custom repository. Supports private
# gcr.io images if the Composer Environment is under the same
# project-id as the gcr.io images and the service account that Composer
# uses has permission to access the Google Container Registry
# (the default service account has permission)
image="gcr.io/gcp-runtimes/ubuntu_18_0_4",
)
kubenetes_template_ex = KubernetesPodOperator(
task_id="ex-kube-templates",
name="ex-kube-templates",
namespace="default",
image="bash",
# All parameters below are able to be templated with jinja -- cmds,
# arguments, env_vars, and config_file. For more information visit:
# https://airflow.apache.org/docs/apache-airflow/stable/macros-ref.html
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# DS in jinja is the execution date as YYYY-MM-DD, this docker image
# will echo the execution date. Arguments to the entrypoint. The docker
# image's CMD is used if this is not provided. The arguments parameter
# is templated.
arguments=["{{ ds }}"],
# The var template variable allows you to access variables defined in
# Airflow UI. In this case we are getting the value of my_value and
# setting the environment variable `MY_VALUE`. The pod will fail if
# `my_value` is not set in the Airflow UI.
env_vars={"MY_VALUE": "{{ var.value.my_value }}"},
# Sets the config file to a kubernetes config file specified in
# airflow.cfg. If the configuration file does not exist or does
# not provide validcredentials the pod will fail to launch. If not
# specified, config_file defaults to ~/.kube/config
config_file="{{ conf.get('core', 'kube_config') }}",
)
kubernetes_secret_vars_ex = KubernetesPodOperator(
task_id="ex-kube-secrets",
name="ex-kube-secrets",
namespace="default",
image="ubuntu",
startup_timeout_seconds=300,
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[secret_env, secret_volume],
# env_vars allows you to specify environment variables for your
# container to use. env_vars is templated.
env_vars={
"EXAMPLE_VAR": "/example/value",
"GOOGLE_APPLICATION_CREDENTIALS": "/var/secrets/google/service-account.json ",
},
)
# Pod affinity with the KubernetesPodOperator
# is not supported with Composer 2
# instead, create a cluster and use the GKEStartPodOperator
# https://cloud.google.com/composer/docs/using-gke-operator
kubernetes_affinity_ex = KubernetesPodOperator(
task_id="ex-pod-affinity",
name="ex-pod-affinity",
namespace="default",
image="perl:5.34.0",
cmds=["perl"],
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# affinity allows you to constrain which nodes your pod is eligible to
# be scheduled on, based on labels on the node. In this case, if the
# label 'cloud.google.com/gke-nodepool' with value
# 'nodepool-label-value' or 'nodepool-label-value2' is not found on any
# nodes, it will fail to schedule.
affinity={
"nodeAffinity": {
# requiredDuringSchedulingIgnoredDuringExecution means in order
# for a pod to be scheduled on a node, the node must have the
# specified labels. However, if labels on a node change at
# runtime such that the affinity rules on a pod are no longer
# met, the pod will still continue to run on the node.
"requiredDuringSchedulingIgnoredDuringExecution": {
"nodeSelectorTerms": [
{
"matchExpressions": [
{
# When nodepools are created in Google Kubernetes
# Engine, the nodes inside of that nodepool are
# automatically assigned the label
# 'cloud.google.com/gke-nodepool' with the value of
# the nodepool's name.
"key": "cloud.google.com/gke-nodepool",
"operator": "In",
# The label key's value that pods can be scheduled
# on.
"values": [
"pool-0",
"pool-1",
],
}
]
}
]
}
}
},
)
kubernetes_full_pod = KubernetesPodOperator(
task_id="ex-all-configs",
name="pi",
namespace="default",
image="perl:5.34.0",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["perl"],
# Arguments to the entrypoint. The docker image's CMD is used if this
# is not provided. The arguments parameter is templated.
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[],
# Labels to apply to the Pod.
labels={"pod-label": "label-name"},
# Timeout to start up the Pod, default is 120.
startup_timeout_seconds=120,
# The environment variables to be initialized in the container
# env_vars are templated.
env_vars={"EXAMPLE_VAR": "/example/value"},
# If true, logs stdout output of container. Defaults to True.
get_logs=True,
# Determines when to pull a fresh image, if 'IfNotPresent' will cause
# the Kubelet to skip pulling an image if it already exists. If you
# want to always pull a new image, set it to 'Always'.
image_pull_policy="Always",
# Annotations are non-identifying metadata you can attach to the Pod.
# Can be a large range of data, and can include characters that are not
# permitted by labels.
annotations={"key1": "value1"},
# Optional resource specifications for Pod, this will allow you to
# set both cpu and memory limits and requirements.
# Prior to Airflow 2.3 and the cncf providers package 5.0.0
# resources were passed as a dictionary. This change was made in
# https://github.com/apache/airflow/pull/27197
# Additionally, "memory" and "cpu" were previously named
# "limit_memory" and "limit_cpu"
# resources={'limit_memory': "250M", 'limit_cpu': "100m"},
container_resources=k8s_models.V1ResourceRequirements(
limits={"memory": "250M", "cpu": "100m"},
),
# Specifies path to kubernetes config. If no config is specified will
# default to '~/.kube/config'. The config_file is templated.
config_file="/home/airflow/composer_kube_config",
# If true, the content of /airflow/xcom/return.json from container will
# also be pushed to an XCom when the container ends.
do_xcom_push=False,
# List of Volume objects to pass to the Pod.
volumes=[],
# List of VolumeMount objects to pass to the Pod.
volume_mounts=[],
# Affinity determines which nodes the Pod can run on based on the
# config. For more information see:
# https://kubernetes.io/docs/concepts/configuration/assign-pod-node/
# Pod affinity with the KubernetesPodOperator
# is not supported with Composer 2
# instead, create a cluster and use the GKEStartPodOperator
# https://cloud.google.com/composer/docs/using-gke-operator
affinity={},
)
import datetime
from airflow import models
from airflow.contrib.kubernetes import secret
from airflow.contrib.operators import kubernetes_pod_operator
# A Secret is an object that contains a small amount of sensitive data such as
# a password, a token, or a key. Such information might otherwise be put in a
# Pod specification or in an image; putting it in a Secret object allows for
# more control over how it is used, and reduces the risk of accidental
# exposure.
secret_env = secret.Secret(
# Expose the secret as environment variable.
deploy_type="env",
# The name of the environment variable, since deploy_type is `env` rather
# than `volume`.
deploy_target="SQL_CONN",
# Name of the Kubernetes Secret
secret="airflow-secrets",
# Key of a secret stored in this Secret object
key="sql_alchemy_conn",
)
secret_volume = secret.Secret(
deploy_type="volume",
# Path where we mount the secret as volume
deploy_target="/var/secrets/google",
# Name of Kubernetes Secret
secret="service-account",
# Key in the form of service account file name
key="service-account.json",
)
# If you are running Airflow in more than one time zone
# see https://airflow.apache.org/docs/apache-airflow/stable/timezone.html
# for best practices
YESTERDAY = datetime.datetime.now() - datetime.timedelta(days=1)
# If a Pod fails to launch, or has an error occur in the container, Airflow
# will show the task as failed, as well as contain all of the task logs
# required to debug.
with models.DAG(
dag_id="composer_sample_kubernetes_pod",
schedule_interval=datetime.timedelta(days=1),
start_date=YESTERDAY,
) as dag:
# Only name, namespace, image, and task_id are required to create a
# KubernetesPodOperator. In Cloud Composer, currently the operator defaults
# to using the config file found at `/home/airflow/composer_kube_config if
# no `config_file` parameter is specified. By default it will contain the
# credentials for Cloud Composer's Google Kubernetes Engine cluster that is
# created upon environment creation.
kubernetes_min_pod = kubernetes_pod_operator.KubernetesPodOperator(
# The ID specified for the task.
task_id="pod-ex-minimum",
# Name of task you want to run, used to generate Pod ID.
name="pod-ex-minimum",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# The namespace to run within Kubernetes, default namespace is
# `default`. There is the potential for the resource starvation of
# Airflow workers and scheduler within the Cloud Composer environment,
# the recommended solution is to increase the amount of nodes in order
# to satisfy the computing requirements. Alternatively, launching pods
# into a custom namespace will stop fighting over resources.
namespace="default",
# Docker image specified. Defaults to hub.docker.com, but any fully
# qualified URLs will point to a custom repository. Supports private
# gcr.io images if the Composer Environment is under the same
# project-id as the gcr.io images and the service account that Composer
# uses has permission to access the Google Container Registry
# (the default service account has permission)
image="gcr.io/gcp-runtimes/ubuntu_18_0_4",
)
kubenetes_template_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-kube-templates",
name="ex-kube-templates",
namespace="default",
image="bash",
# All parameters below are able to be templated with jinja -- cmds,
# arguments, env_vars, and config_file. For more information visit:
# https://airflow.apache.org/docs/apache-airflow/stable/macros-ref.html
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# DS in jinja is the execution date as YYYY-MM-DD, this docker image
# will echo the execution date. Arguments to the entrypoint. The docker
# image's CMD is used if this is not provided. The arguments parameter
# is templated.
arguments=["{{ ds }}"],
# The var template variable allows you to access variables defined in
# Airflow UI. In this case we are getting the value of my_value and
# setting the environment variable `MY_VALUE`. The pod will fail if
# `my_value` is not set in the Airflow UI.
env_vars={"MY_VALUE": "{{ var.value.my_value }}"},
# Sets the config file to a kubernetes config file specified in
# airflow.cfg. If the configuration file does not exist or does
# not provide validcredentials the pod will fail to launch. If not
# specified, config_file defaults to ~/.kube/config
config_file="{{ conf.get('core', 'kube_config') }}",
)
kubernetes_secret_vars_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-kube-secrets",
name="ex-kube-secrets",
namespace="default",
image="ubuntu",
startup_timeout_seconds=300,
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[secret_env, secret_volume],
# env_vars allows you to specify environment variables for your
# container to use. env_vars is templated.
env_vars={
"EXAMPLE_VAR": "/example/value",
"GOOGLE_APPLICATION_CREDENTIALS": "/var/secrets/google/service-account.json ",
},
)
kubernetes_affinity_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-pod-affinity",
name="ex-pod-affinity",
namespace="default",
image="perl:5.34.0",
cmds=["perl"],
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# affinity allows you to constrain which nodes your pod is eligible to
# be scheduled on, based on labels on the node. In this case, if the
# label 'cloud.google.com/gke-nodepool' with value
# 'nodepool-label-value' or 'nodepool-label-value2' is not found on any
# nodes, it will fail to schedule.
affinity={
"nodeAffinity": {
# requiredDuringSchedulingIgnoredDuringExecution means in order
# for a pod to be scheduled on a node, the node must have the
# specified labels. However, if labels on a node change at
# runtime such that the affinity rules on a pod are no longer
# met, the pod will still continue to run on the node.
"requiredDuringSchedulingIgnoredDuringExecution": {
"nodeSelectorTerms": [
{
"matchExpressions": [
{
# When nodepools are created in Google Kubernetes
# Engine, the nodes inside of that nodepool are
# automatically assigned the label
# 'cloud.google.com/gke-nodepool' with the value of
# the nodepool's name.
"key": "cloud.google.com/gke-nodepool",
"operator": "In",
# The label key's value that pods can be scheduled
# on.
"values": [
"pool-0",
"pool-1",
],
}
]
}
]
}
}
},
)
kubernetes_full_pod = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-all-configs",
name="pi",
namespace="default",
image="perl:5.34.0",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["perl"],
# Arguments to the entrypoint. The docker image's CMD is used if this
# is not provided. The arguments parameter is templated.
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[],
# Labels to apply to the Pod.
labels={"pod-label": "label-name"},
# Timeout to start up the Pod, default is 120.
startup_timeout_seconds=120,
# The environment variables to be initialized in the container
# env_vars are templated.
env_vars={"EXAMPLE_VAR": "/example/value"},
# If true, logs stdout output of container. Defaults to True.
get_logs=True,
# Determines when to pull a fresh image, if 'IfNotPresent' will cause
# the Kubelet to skip pulling an image if it already exists. If you
# want to always pull a new image, set it to 'Always'.
image_pull_policy="Always",
# Annotations are non-identifying metadata you can attach to the Pod.
# Can be a large range of data, and can include characters that are not
# permitted by labels.
annotations={"key1": "value1"},
# Optional resource specifications for Pod, this will allow you to
# set both cpu and memory limits and requirements.
# Prior to Airflow 1.10.4, resource specifications were
# passed as a Pod Resources Class object,
# If using this example on a version of Airflow prior to 1.10.4,
# import the "pod" package from airflow.contrib.kubernetes and use
# resources = pod.Resources() instead passing a dict
# For more info see:
# https://github.com/apache/airflow/pull/4551
resources={"limit_memory": "250M", "limit_cpu": "100m"},
# Specifies path to kubernetes config. If no config is specified will
# default to '~/.kube/config'. The config_file is templated.
config_file="/home/airflow/composer_kube_config",
# If true, the content of /airflow/xcom/return.json from container will
# also be pushed to an XCom when the container ends.
do_xcom_push=False,
# List of Volume objects to pass to the Pod.
volumes=[],
# List of VolumeMount objects to pass to the Pod.
volume_mounts=[],
# Affinity determines which nodes the Pod can run on based on the
# config. For more information see:
# https://kubernetes.io/docs/concepts/configuration/assign-pod-node/
affinity={},
)
Configuración mínima
Para crear un KubernetesPodOperator, solo se requieren el name del Pod, el namespace en el que se ejecutará el Pod, los image para usarlo y el task_id.
Cuando colocas el siguiente fragmento de código en un DAG, la configuración utiliza los valores predeterminados en /home/airflow/composer_kube_config. No es necesario modificar el código para que la tarea pod-ex-minimum se realice correctamente.
kubernetes_min_pod = KubernetesPodOperator(
# The ID specified for the task.
task_id="pod-ex-minimum",
# Name of task you want to run, used to generate Pod ID.
name="pod-ex-minimum",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# The namespace to run within Kubernetes, default namespace is
# `default`. In Composer 1 there is the potential for
# the resource starvation of Airflow workers and scheduler
# within the Cloud Composer environment,
# the recommended solution is to increase the amount of nodes in order
# to satisfy the computing requirements. Alternatively, launching pods
# into a custom namespace will stop fighting over resources,
# and using Composer 2 will mean the environment will autoscale.
namespace="default",
# Docker image specified. Defaults to hub.docker.com, but any fully
# qualified URLs will point to a custom repository. Supports private
# gcr.io images if the Composer Environment is under the same
# project-id as the gcr.io images and the service account that Composer
# uses has permission to access the Google Container Registry
# (the default service account has permission)
image="gcr.io/gcp-runtimes/ubuntu_18_0_4",
)
kubernetes_min_pod = kubernetes_pod_operator.KubernetesPodOperator(
# The ID specified for the task.
task_id="pod-ex-minimum",
# Name of task you want to run, used to generate Pod ID.
name="pod-ex-minimum",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# The namespace to run within Kubernetes, default namespace is
# `default`. There is the potential for the resource starvation of
# Airflow workers and scheduler within the Cloud Composer environment,
# the recommended solution is to increase the amount of nodes in order
# to satisfy the computing requirements. Alternatively, launching pods
# into a custom namespace will stop fighting over resources.
namespace="default",
# Docker image specified. Defaults to hub.docker.com, but any fully
# qualified URLs will point to a custom repository. Supports private
# gcr.io images if the Composer Environment is under the same
# project-id as the gcr.io images and the service account that Composer
# uses has permission to access the Google Container Registry
# (the default service account has permission)
image="gcr.io/gcp-runtimes/ubuntu_18_0_4",
)
Configuración de la plantilla
Airflow admite el uso de plantillas de Jinja.
Debes declarar las variables obligatorias (task_id, name, namespace y image) con el operador. Como se muestra en el siguiente ejemplo, puedes crear plantillas de todos los demás parámetros con Jinja, incluidos cmds, arguments, env_vars y config_file.
kubenetes_template_ex = KubernetesPodOperator(
task_id="ex-kube-templates",
name="ex-kube-templates",
namespace="default",
image="bash",
# All parameters below are able to be templated with jinja -- cmds,
# arguments, env_vars, and config_file. For more information visit:
# https://airflow.apache.org/docs/apache-airflow/stable/macros-ref.html
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# DS in jinja is the execution date as YYYY-MM-DD, this docker image
# will echo the execution date. Arguments to the entrypoint. The docker
# image's CMD is used if this is not provided. The arguments parameter
# is templated.
arguments=["{{ ds }}"],
# The var template variable allows you to access variables defined in
# Airflow UI. In this case we are getting the value of my_value and
# setting the environment variable `MY_VALUE`. The pod will fail if
# `my_value` is not set in the Airflow UI.
env_vars={"MY_VALUE": "{{ var.value.my_value }}"},
# Sets the config file to a kubernetes config file specified in
# airflow.cfg. If the configuration file does not exist or does
# not provide validcredentials the pod will fail to launch. If not
# specified, config_file defaults to ~/.kube/config
config_file="{{ conf.get('core', 'kube_config') }}",
)
kubenetes_template_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-kube-templates",
name="ex-kube-templates",
namespace="default",
image="bash",
# All parameters below are able to be templated with jinja -- cmds,
# arguments, env_vars, and config_file. For more information visit:
# https://airflow.apache.org/docs/apache-airflow/stable/macros-ref.html
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["echo"],
# DS in jinja is the execution date as YYYY-MM-DD, this docker image
# will echo the execution date. Arguments to the entrypoint. The docker
# image's CMD is used if this is not provided. The arguments parameter
# is templated.
arguments=["{{ ds }}"],
# The var template variable allows you to access variables defined in
# Airflow UI. In this case we are getting the value of my_value and
# setting the environment variable `MY_VALUE`. The pod will fail if
# `my_value` is not set in the Airflow UI.
env_vars={"MY_VALUE": "{{ var.value.my_value }}"},
# Sets the config file to a kubernetes config file specified in
# airflow.cfg. If the configuration file does not exist or does
# not provide validcredentials the pod will fail to launch. If not
# specified, config_file defaults to ~/.kube/config
config_file="{{ conf.get('core', 'kube_config') }}",
)
Sin cambiar el DAG ni tu entorno, la tarea ex-kube-templates falla debido a dos errores. Los registros muestran que esta tarea falla porque no existe la variable adecuada (my_value). El segundo error, que puedes obtener después de corregir el primer error, muestra que la tarea falla porque core/kube_config no se encuentra en config.
Para corregir ambos errores, sigue los pasos que se describen en detalle.
Para configurar my_value con gcloud o la IU de Airflow:
LOCATION por la región en la que se encuentra el entorno
Para hacer referencia a un config_file personalizado (un archivo de configuración de Kubernetes), anula la opción de configuración de Airflow kube_config para una configuración válida de Kubernetes:
Sección
Clave
Valor
core
kube_config
/home/airflow/composer_kube_config
Espera unos minutos a que se actualice tu entorno. Luego, vuelve a ejecutar la tarea ex-kube-templates y verifica que la tarea ex-kube-templates se complete con éxito.
Configuración de variables secretas
Un secreto de Kubernetes es un objeto que contiene datos sensibles. Puedes pasar secretos a los pods de Kubernetes mediante KubernetesPodOperator.
Los secretos deben estar definidos en Kubernetes o el pod no se iniciará.
En este ejemplo, se muestran dos formas de usar los Secrets de Kubernetes: como una variable de entorno y como un volumen activado por el Pod.
El primer secreto, airflow-secrets, se establece en una variable de entorno de Kubernetes llamada SQL_CONN (en lugar de en una variable de entorno de Airflow o Cloud Composer).
El segundo secreto, service-account, activa service-account.json, un archivo con un token de cuenta de servicio, en /var/secrets/google.
secret_env = Secret(
# Expose the secret as environment variable.
deploy_type="env",
# The name of the environment variable, since deploy_type is `env` rather
# than `volume`.
deploy_target="SQL_CONN",
# Name of the Kubernetes Secret
secret="airflow-secrets",
# Key of a secret stored in this Secret object
key="sql_alchemy_conn",
)
secret_volume = Secret(
deploy_type="volume",
# Path where we mount the secret as volume
deploy_target="/var/secrets/google",
# Name of Kubernetes Secret
secret="service-account",
# Key in the form of service account file name
key="service-account.json",
)
secret_env = secret.Secret(
# Expose the secret as environment variable.
deploy_type="env",
# The name of the environment variable, since deploy_type is `env` rather
# than `volume`.
deploy_target="SQL_CONN",
# Name of the Kubernetes Secret
secret="airflow-secrets",
# Key of a secret stored in this Secret object
key="sql_alchemy_conn",
)
secret_volume = secret.Secret(
deploy_type="volume",
# Path where we mount the secret as volume
deploy_target="/var/secrets/google",
# Name of Kubernetes Secret
secret="service-account",
# Key in the form of service account file name
key="service-account.json",
)
El nombre del primer secreto de Kubernetes se define en la variable secret.
Este secreto específico se llama airflow-secrets. Se expone como una variable de entorno, según lo determina el deploy_type. La variable de entorno que establece, deploy_target, es SQL_CONN. Por último, el key del secreto que se almacena en deploy_target es sql_alchemy_conn.
El nombre del segundo secreto de Kubernetes se define en la variable secret.
Este secreto específico se llama service-account. Se expone como un volumen, según lo determina el deploy_type. La ruta de acceso del archivo para activar, deploy_target, es /var/secrets/google. Por último, el key del secreto almacenado en deploy_target es service-account.json.
kubernetes_secret_vars_ex = KubernetesPodOperator(
task_id="ex-kube-secrets",
name="ex-kube-secrets",
namespace="default",
image="ubuntu",
startup_timeout_seconds=300,
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[secret_env, secret_volume],
# env_vars allows you to specify environment variables for your
# container to use. env_vars is templated.
env_vars={
"EXAMPLE_VAR": "/example/value",
"GOOGLE_APPLICATION_CREDENTIALS": "/var/secrets/google/service-account.json ",
},
)
kubernetes_secret_vars_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-kube-secrets",
name="ex-kube-secrets",
namespace="default",
image="ubuntu",
startup_timeout_seconds=300,
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[secret_env, secret_volume],
# env_vars allows you to specify environment variables for your
# container to use. env_vars is templated.
env_vars={
"EXAMPLE_VAR": "/example/value",
"GOOGLE_APPLICATION_CREDENTIALS": "/var/secrets/google/service-account.json ",
},
)
Si no realizas ningún cambio en el DAG ni en tu entorno, la tarea ex-kube-secrets falla. Si observas los registros, la tarea falla debido a un error Pod took too long to start. Este error se produce porque Airflow no puede encontrar el secreto especificado en la configuración, secret_env.
gcloud
Para configurar el secreto con gcloud, haz lo siguiente:
Obtén información sobre tu clúster de entorno de Cloud Composer.
Ejecuta el siguiente comando para crear un secreto de Kubernetes que establezca el valor de service-account.json en una ruta local del archivo de claves de una cuenta de servicio llamado key.json:
Después de configurar los secretos, vuelve a ejecutar la tarea ex-kube-secrets en la IU de Airflow.
Verifica que la tarea ex-kube-secrets se realice correctamente.
Configuración de afinidad de Pods
Si configuras el parámetro affinity en el KubernetesPodOperator, puedes controlar en qué nodos se programan los pods (por ejemplo, puedes especificar un grupo de nodos en particular). En este ejemplo, el operador solo se ejecuta en grupos de nodos llamados pool-0 y pool-1. Los nodos de tu entorno de Cloud Composer 1 se encuentran en default-pool, por lo que los Pods no se ejecutan en los nodos de tu entorno.
# Pod affinity with the KubernetesPodOperator
# is not supported with Composer 2
# instead, create a cluster and use the GKEStartPodOperator
# https://cloud.google.com/composer/docs/using-gke-operator
kubernetes_affinity_ex = KubernetesPodOperator(
task_id="ex-pod-affinity",
name="ex-pod-affinity",
namespace="default",
image="perl:5.34.0",
cmds=["perl"],
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# affinity allows you to constrain which nodes your pod is eligible to
# be scheduled on, based on labels on the node. In this case, if the
# label 'cloud.google.com/gke-nodepool' with value
# 'nodepool-label-value' or 'nodepool-label-value2' is not found on any
# nodes, it will fail to schedule.
affinity={
"nodeAffinity": {
# requiredDuringSchedulingIgnoredDuringExecution means in order
# for a pod to be scheduled on a node, the node must have the
# specified labels. However, if labels on a node change at
# runtime such that the affinity rules on a pod are no longer
# met, the pod will still continue to run on the node.
"requiredDuringSchedulingIgnoredDuringExecution": {
"nodeSelectorTerms": [
{
"matchExpressions": [
{
# When nodepools are created in Google Kubernetes
# Engine, the nodes inside of that nodepool are
# automatically assigned the label
# 'cloud.google.com/gke-nodepool' with the value of
# the nodepool's name.
"key": "cloud.google.com/gke-nodepool",
"operator": "In",
# The label key's value that pods can be scheduled
# on.
"values": [
"pool-0",
"pool-1",
],
}
]
}
]
}
}
},
)
kubernetes_affinity_ex = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-pod-affinity",
name="ex-pod-affinity",
namespace="default",
image="perl:5.34.0",
cmds=["perl"],
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# affinity allows you to constrain which nodes your pod is eligible to
# be scheduled on, based on labels on the node. In this case, if the
# label 'cloud.google.com/gke-nodepool' with value
# 'nodepool-label-value' or 'nodepool-label-value2' is not found on any
# nodes, it will fail to schedule.
affinity={
"nodeAffinity": {
# requiredDuringSchedulingIgnoredDuringExecution means in order
# for a pod to be scheduled on a node, the node must have the
# specified labels. However, if labels on a node change at
# runtime such that the affinity rules on a pod are no longer
# met, the pod will still continue to run on the node.
"requiredDuringSchedulingIgnoredDuringExecution": {
"nodeSelectorTerms": [
{
"matchExpressions": [
{
# When nodepools are created in Google Kubernetes
# Engine, the nodes inside of that nodepool are
# automatically assigned the label
# 'cloud.google.com/gke-nodepool' with the value of
# the nodepool's name.
"key": "cloud.google.com/gke-nodepool",
"operator": "In",
# The label key's value that pods can be scheduled
# on.
"values": [
"pool-0",
"pool-1",
],
}
]
}
]
}
}
},
)
Con la configuración actual de este ejemplo, la tarea falla. Si observas los registros, la tarea falla porque los grupos de nodos pool-0 y pool-1 no existen.
Para asegurarte de que los grupos de nodos de values existan, realiza cualquiera de los siguientes cambios de configuración:
Si creaste un grupo de nodos anteriormente, reemplaza pool-0 y pool-1 con los nombres de tus grupos de nodos y vuelve a subir el DAG.
Crea un grupo de nodos con el nombre pool-0 o pool-1. Puedes crear ambos, pero la tarea necesita solo uno para tener éxito.
Reemplaza pool-0 y pool-1 con default-pool, que es el grupo predeterminado que utiliza Airflow. A continuación, vuelve a subir el DAG.
Después de realizar los cambios, espera unos minutos hasta que se actualice tu entorno.
Luego, vuelve a ejecutar la tarea ex-pod-affinity y verifica que la tarea ex-pod-affinity se complete correctamente.
Configuración completa
En este ejemplo, se muestran todas las variables que puedes configurar en KubernetesPodOperator. No es necesario modificar el código para que la tarea ex-all-configs se realice correctamente.
kubernetes_full_pod = KubernetesPodOperator(
task_id="ex-all-configs",
name="pi",
namespace="default",
image="perl:5.34.0",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["perl"],
# Arguments to the entrypoint. The docker image's CMD is used if this
# is not provided. The arguments parameter is templated.
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[],
# Labels to apply to the Pod.
labels={"pod-label": "label-name"},
# Timeout to start up the Pod, default is 120.
startup_timeout_seconds=120,
# The environment variables to be initialized in the container
# env_vars are templated.
env_vars={"EXAMPLE_VAR": "/example/value"},
# If true, logs stdout output of container. Defaults to True.
get_logs=True,
# Determines when to pull a fresh image, if 'IfNotPresent' will cause
# the Kubelet to skip pulling an image if it already exists. If you
# want to always pull a new image, set it to 'Always'.
image_pull_policy="Always",
# Annotations are non-identifying metadata you can attach to the Pod.
# Can be a large range of data, and can include characters that are not
# permitted by labels.
annotations={"key1": "value1"},
# Optional resource specifications for Pod, this will allow you to
# set both cpu and memory limits and requirements.
# Prior to Airflow 2.3 and the cncf providers package 5.0.0
# resources were passed as a dictionary. This change was made in
# https://github.com/apache/airflow/pull/27197
# Additionally, "memory" and "cpu" were previously named
# "limit_memory" and "limit_cpu"
# resources={'limit_memory': "250M", 'limit_cpu': "100m"},
container_resources=k8s_models.V1ResourceRequirements(
limits={"memory": "250M", "cpu": "100m"},
),
# Specifies path to kubernetes config. If no config is specified will
# default to '~/.kube/config'. The config_file is templated.
config_file="/home/airflow/composer_kube_config",
# If true, the content of /airflow/xcom/return.json from container will
# also be pushed to an XCom when the container ends.
do_xcom_push=False,
# List of Volume objects to pass to the Pod.
volumes=[],
# List of VolumeMount objects to pass to the Pod.
volume_mounts=[],
# Affinity determines which nodes the Pod can run on based on the
# config. For more information see:
# https://kubernetes.io/docs/concepts/configuration/assign-pod-node/
# Pod affinity with the KubernetesPodOperator
# is not supported with Composer 2
# instead, create a cluster and use the GKEStartPodOperator
# https://cloud.google.com/composer/docs/using-gke-operator
affinity={},
)
kubernetes_full_pod = kubernetes_pod_operator.KubernetesPodOperator(
task_id="ex-all-configs",
name="pi",
namespace="default",
image="perl:5.34.0",
# Entrypoint of the container, if not specified the Docker container's
# entrypoint is used. The cmds parameter is templated.
cmds=["perl"],
# Arguments to the entrypoint. The docker image's CMD is used if this
# is not provided. The arguments parameter is templated.
arguments=["-Mbignum=bpi", "-wle", "print bpi(2000)"],
# The secrets to pass to Pod, the Pod will fail to create if the
# secrets you specify in a Secret object do not exist in Kubernetes.
secrets=[],
# Labels to apply to the Pod.
labels={"pod-label": "label-name"},
# Timeout to start up the Pod, default is 120.
startup_timeout_seconds=120,
# The environment variables to be initialized in the container
# env_vars are templated.
env_vars={"EXAMPLE_VAR": "/example/value"},
# If true, logs stdout output of container. Defaults to True.
get_logs=True,
# Determines when to pull a fresh image, if 'IfNotPresent' will cause
# the Kubelet to skip pulling an image if it already exists. If you
# want to always pull a new image, set it to 'Always'.
image_pull_policy="Always",
# Annotations are non-identifying metadata you can attach to the Pod.
# Can be a large range of data, and can include characters that are not
# permitted by labels.
annotations={"key1": "value1"},
# Optional resource specifications for Pod, this will allow you to
# set both cpu and memory limits and requirements.
# Prior to Airflow 1.10.4, resource specifications were
# passed as a Pod Resources Class object,
# If using this example on a version of Airflow prior to 1.10.4,
# import the "pod" package from airflow.contrib.kubernetes and use
# resources = pod.Resources() instead passing a dict
# For more info see:
# https://github.com/apache/airflow/pull/4551
resources={"limit_memory": "250M", "limit_cpu": "100m"},
# Specifies path to kubernetes config. If no config is specified will
# default to '~/.kube/config'. The config_file is templated.
config_file="/home/airflow/composer_kube_config",
# If true, the content of /airflow/xcom/return.json from container will
# also be pushed to an XCom when the container ends.
do_xcom_push=False,
# List of Volume objects to pass to the Pod.
volumes=[],
# List of VolumeMount objects to pass to the Pod.
volume_mounts=[],
# Affinity determines which nodes the Pod can run on based on the
# config. For more information see:
# https://kubernetes.io/docs/concepts/configuration/assign-pod-node/
affinity={},
)
Información sobre el proveedor de Kubernetes de CNCF
GKEStartPodOperator y KubernetesPodOperator se implementan dentro del proveedor apache-airflow-providers-cncf-kubernetes.
En la versión 6.0.0 del paquete del proveedor de Kubernetes de CNCF, la conexión kubernetes_default se usa de forma predeterminada en KubernetesPodOperator.
Si especificaste una conexión personalizada en la versión 5.0.0, el operador aún la usará. Para volver a usar la conexión kubernetes_default, es posible que desees ajustar tus DAG según corresponda.
Versión 5.0.0
Esta versión presenta algunos cambios incompatibles con versiones anteriores en comparación con la versión 4.4.0. Los más importantes, que debes tener en cuenta, están relacionados con la conexión kubernetes_default, que no se usa en la versión 5.0.0.
La conexión kubernetes_default debe modificarse: la ruta de configuración de Kube debe establecerse en /home/airflow/composer_kube_config o, de manera alternativa, se debe agregar config_file a la configuración KubernetesPodOperator (como se presentó a continuación).
Modifica el código de una tarea con KubernetesPodOperator de la siguiente manera.
KubernetesPodOperator(
# config_file parameter - can be skipped if connection contains this setting
config_file="/home/airflow/composer_kube_config",
# definition of connection to be used by the operator
kubernetes_conn_id='kubernetes_default',
...
)
Revisa los registros de la carpeta logs/<DAG_NAME>/<TASK_ID>/<EXECUTION_DATE>.
Registros detallados del Pod en la consola de Google Cloud, en las cargas de trabajo de GKE. Estos registros incluyen el archivo YAML de definición de pod, los eventos de los pods y sus detalles.
Códigos de retorno distintos de cero cuando también se usa el GKEStartPodOperator
Cuando se usa KubernetesPodOperator y GKEStartPodOperator, el código de retorno del punto de entrada del contenedor determina si la tarea se considera exitosa o no. Los códigos de retorno distintos de cero indican un error.
Un patrón común cuando se utiliza KubernetesPodOperator y GKEStartPodOperator es ejecutar una secuencia de comandos de shell como punto de entrada de contenedor para agrupar varias operaciones dentro de este.
Si escribes una secuencia de comandos de este tipo, recomendamos que incluyas el comando set -e en la parte superior de la secuencia de comandos para que sus comandos con error finalicen la secuencia y propaguen el error a la instancia de tarea de Airflow.
Tiempos de espera de los pods
El tiempo de espera predeterminado de KubernetesPodOperator es de 120 segundos, lo que puede provocar que el tiempo de espera se agote antes de que se descarguen las imágenes más grandes. Para aumentar el tiempo de espera, puedes modificar el parámetro startup_timeout_seconds cuando creas el KubernetesPodOperator.
Cuando se agota el tiempo de espera de un Pod, el registro específico de la tarea está disponible en la IU de Airflow. Por ejemplo:
Executing <Task(KubernetesPodOperator): ex-all-configs> on 2018-07-23 19:06:58.133811
Running: ['bash', '-c', u'airflow run kubernetes-pod-example ex-all-configs 2018-07-23T19:06:58.133811 --job_id 726 --raw -sd DAGS_FOLDER/kubernetes_pod_operator_sample.py']
Event: pod-name-9a8e9d06 had an event of type Pending
...
...
Event: pod-name-9a8e9d06 had an event of type Pending
Traceback (most recent call last):
File "/usr/local/bin/airflow", line 27, in <module>
args.func(args)
File "/usr/local/lib/python2.7/site-packages/airflow/bin/cli.py", line 392, in run
pool=args.pool,
File "/usr/local/lib/python2.7/site-packages/airflow/utils/db.py", line 50, in wrapper
result = func(*args, **kwargs)
File "/usr/local/lib/python2.7/site-packages/airflow/models.py", line 1492, in _run_raw_task
result = task_copy.execute(context=context)
File "/usr/local/lib/python2.7/site-packages/airflow/contrib/operators/kubernetes_pod_operator.py", line 123, in execute
raise AirflowException('Pod Launching failed: {error}'.format(error=ex))
airflow.exceptions.AirflowException: Pod Launching failed: Pod took too long to start
También es posible que se agote el tiempo de espera de los pods cuando la cuenta de servicio de Composer carece de los permisos de IAM necesarios para realizar la tarea solicitada. Si deseas verificar esto, revisa los errores en el nivel del pod mediante los Paneles de GKE para ver los registros de tu carga de trabajo específica o usa Cloud Logging.
No se pudo establecer una nueva conexión
La actualización automática está habilitada de forma predeterminada en los clústeres de GKE.
Si un grupo de nodos está en un clúster que se está actualizando, es posible que veas el siguiente error:
<Task(KubernetesPodOperator): gke-upgrade> Failed to establish a new
connection: [Errno 111] Connection refused
Para verificar si tu clúster se está actualizando, en la consola de Google Cloud, ve a la página
Clústeres de Kubernetes y busca el ícono de carga junto al
nombre del clúster de tu entorno.
