Utilizzo degli operatori GKE

Cloud Composer 3 | Cloud Composer 2 | Cloud Composer 1

Nota: questa pagina non è ancora stata rivista per Cloud Composer 3 e mostra i contenuti per Cloud Composer 2.

Questa pagina descrive come utilizzare gli operatori Google Kubernetes Engine per creare cluster in Google Kubernetes Engine e per avviare pod Kubernetes in questi cluster.

Gli operatori Google Kubernetes Engine eseguono i pod Kubernetes in un cluster specificato, che può essere un cluster separato non correlato al tuo ambiente. Al contrario, KubernetesPodOperator esegue i pod Kubernetes nel cluster del tuo ambiente.

Questa pagina descrive un DAG di esempio che crea un cluster Google Kubernetes Engine con GKECreateClusterOperator, utilizza GKEStartPodOperator con le seguenti configurazioni e poi lo elimina con GKEDeleteClusterOperator:

• Configurazione minima: imposta solo i parametri obbligatori.
• Configurazione del modello: utilizza parametri che puoi modificare con Jinja.
• Configurazione dell'affinità pod: limita i nodi disponibili su cui pianificare i pod.
• Configurazione completa: include tutte le configurazioni.

Prima di iniziare

Attenzione: se avvii i pod nel cluster del tuo ambiente, non utilizzare spazi dei nomi Cloud Composer (composer-*) diversi da composer-user-workloads. Questi spazi dei nomi possono cambiare durante le operazioni di upgrade dell'ambiente. Questo potrebbe causare l'interruzione del funzionamento dei DAG.

Configurazione dell'operatore GKE

Per seguire questo esempio, inserisci l'intero file gke_operator.py nella cartella dags/ del tuo ambiente o aggiungi il codice pertinente a un DAG.

Crea un cluster

Il codice mostrato qui crea un cluster Google Kubernetes Engine con due node pool, pool-0 e pool-1, ognuno con un nodo. Se necessario, puoi impostare altri parametri dall'API Google Kubernetes Engine nell'ambito di body.

Ti consigliamo di utilizzare i cluster regionali in Airflow 2. I cluster a livello di zona sono più esposti a errori a livello di zona. Ad esempio, potresti voler utilizzare la regione us-central1 per il cluster anziché la zona us-central1-a. Per saperne di più sulle considerazioni specifiche per le regioni, consulta Area geografica e regioni.

Prima del rilascio della versione 5.1.0 di apache-airflow-providers-google, non era possibile passare l'oggetto node_pools in GKECreateClusterOperator. Se utilizzi Airflow 2, assicurati che il tuo ambiente utilizzi apache-airflow-providers-google versione 5.1.0 o successive. Puoi installare una versione più recente di questo pacchetto PyPI specificando apache-airflow-providers-google e >=5.1.0 come versione richiesta.

# TODO(developer): update with your values
PROJECT_ID = "my-project-id"
# It is recommended to use regional clusters for increased reliability
# though passing a zone in the location parameter is also valid
CLUSTER_REGION = "us-west1"
CLUSTER_NAME = "example-cluster"
CLUSTER = {
    "name": CLUSTER_NAME,
    "node_pools": [
        {"name": "pool-0", "initial_node_count": 1},
        {"name": "pool-1", "initial_node_count": 1},
    ],
}
create_cluster = GKECreateClusterOperator(
    task_id="create_cluster",
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
    body=CLUSTER,
)

Avviare i carichi di lavoro nel cluster

Le sezioni seguenti spiegano ogni GKEStartPodOperatorconfigurazione nell'esempio. Per informazioni su ogni variabile di configurazione, consulta il riferimento di Airflow per gli operatori GKE.

from airflow import models
from airflow.providers.google.cloud.operators.kubernetes_engine import (
    GKECreateClusterOperator,
    GKEDeleteClusterOperator,
    GKEStartPodOperator,
)
from airflow.utils.dates import days_ago

from kubernetes.client import models as k8s_models


with models.DAG(
    "example_gcp_gke",
    schedule_interval=None,  # Override to match your needs
    start_date=days_ago(1),
    tags=["example"],
) as dag:
    # TODO(developer): update with your values
    PROJECT_ID = "my-project-id"
    # It is recommended to use regional clusters for increased reliability
    # though passing a zone in the location parameter is also valid
    CLUSTER_REGION = "us-west1"
    CLUSTER_NAME = "example-cluster"
    CLUSTER = {
        "name": CLUSTER_NAME,
        "node_pools": [
            {"name": "pool-0", "initial_node_count": 1},
            {"name": "pool-1", "initial_node_count": 1},
        ],
    }
    create_cluster = GKECreateClusterOperator(
        task_id="create_cluster",
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
        body=CLUSTER,
    )

    kubernetes_min_pod = GKEStartPodOperator(
        # 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",
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
        cluster_name=CLUSTER_NAME,
        # 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`.
        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 = GKEStartPodOperator(
        task_id="ex-kube-templates",
        name="ex-kube-templates",
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
        cluster_name=CLUSTER_NAME,
        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 }}"},
    )

    kubernetes_affinity_ex = GKEStartPodOperator(
        task_id="ex-pod-affinity",
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
        cluster_name=CLUSTER_NAME,
        name="ex-pod-affinity",
        namespace="default",
        image="perl",
        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-1",
                                    ],
                                }
                            ]
                        }
                    ]
                }
            }
        },
    )
    kubernetes_full_pod = GKEStartPodOperator(
        task_id="ex-all-configs",
        name="full",
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
        cluster_name=CLUSTER_NAME,
        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"},
        ),
        # 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={},
    )
    delete_cluster = GKEDeleteClusterOperator(
        task_id="delete_cluster",
        name=CLUSTER_NAME,
        project_id=PROJECT_ID,
        location=CLUSTER_REGION,
    )

    create_cluster >> kubernetes_min_pod >> delete_cluster
    create_cluster >> kubernetes_full_pod >> delete_cluster
    create_cluster >> kubernetes_affinity_ex >> delete_cluster
    create_cluster >> kubenetes_template_ex >> delete_cluster

Configurazione minima

Per avviare un pod nel cluster GKE con GKEStartPodOperator, sono necessarie solo le opzioni project_id, location, cluster_name, name, namespace, image e task_id.

Quando inserisci il seguente snippet di codice in un DAG, l'attività pod-ex-minimum ha esito positivo a condizione che i parametri elencati in precedenza siano definiti e validi.

# TODO(developer): update with your values
PROJECT_ID = "my-project-id"
# It is recommended to use regional clusters for increased reliability
# though passing a zone in the location parameter is also valid
CLUSTER_REGION = "us-west1"
CLUSTER_NAME = "example-cluster"
kubernetes_min_pod = GKEStartPodOperator(
    # 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",
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
    cluster_name=CLUSTER_NAME,
    # 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`.
    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",
)

Configurazione del modello

Airflow supporta l'utilizzo di modelli Jinja. Devi dichiarare le variabili richieste (task_id, name, namespace e image) con l'operatore. Come mostrato nell'esempio seguente, puoi creare un modello di tutti gli altri parametri con Jinja, inclusi cmds, arguments e env_vars.

Senza modificare il DAG o l'ambiente, l'attività ex-kube-templates non va a buon fine. Imposta una variabile Airflow denominata my_value per fare in modo che questo DAG venga eseguito correttamente.

Per impostare my_value con gcloud o l'interfaccia utente di Airflow:

gcloud

Per Airflow 2, inserisci il seguente comando:

gcloud composer environments run ENVIRONMENT \
    --location LOCATION \
    variables set -- \
    my_value example_value

Sostituisci:

• ENVIRONMENT con il nome dell'ambiente.
• LOCATION con la regione in cui si trova l'ambiente.

UI di Airflow

Nell'UI di Airflow 2:

1. Nella barra degli strumenti, seleziona Admin > Variables.

2. Nella pagina List Variable (Variabile elenco), fai clic su Add a new record (Aggiungi un nuovo record).

3. Nella pagina Aggiungi variabile, inserisci le seguenti informazioni:

   • Chiave:my_value
   • Valore: example_value

4. Fai clic su Salva.

Configurazione del modello:

# TODO(developer): update with your values
PROJECT_ID = "my-project-id"
# It is recommended to use regional clusters for increased reliability
# though passing a zone in the location parameter is also valid
CLUSTER_REGION = "us-west1"
CLUSTER_NAME = "example-cluster"
kubenetes_template_ex = GKEStartPodOperator(
    task_id="ex-kube-templates",
    name="ex-kube-templates",
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
    cluster_name=CLUSTER_NAME,
    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 }}"},
)

Configurazione dell'affinità dei pod

Quando configuri il parametro affinity in GKEStartPodOperator, controlli su quali nodi programmare i pod, ad esempio i nodi solo in un determinato pool di nodi. Quando hai creato il cluster, hai creato due pool di nodi denominati pool-0 e pool-1. Questo operatore stabilisce che i pod devono essere eseguiti solo in pool-1.

Posizione di avvio del pod Kubernetes di Cloud Composer con affinità pod (fai clic per ingrandire)

# TODO(developer): update with your values
PROJECT_ID = "my-project-id"
# It is recommended to use regional clusters for increased reliability
# though passing a zone in the location parameter is also valid
CLUSTER_REGION = "us-west1"
CLUSTER_NAME = "example-cluster"
kubernetes_affinity_ex = GKEStartPodOperator(
    task_id="ex-pod-affinity",
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
    cluster_name=CLUSTER_NAME,
    name="ex-pod-affinity",
    namespace="default",
    image="perl",
    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-1",
                                ],
                            }
                        ]
                    }
                ]
            }
        }
    },
)

Configurazione completa

Questo esempio mostra tutte le variabili che puoi configurare in GKEStartPodOperator. Non è necessario modificare il codice per la riuscita dell'attività ex-all-configs.

Per informazioni dettagliate su ogni variabile, consulta il riferimento di Airflow per gli operatori GKE.

# TODO(developer): update with your values
PROJECT_ID = "my-project-id"
# It is recommended to use regional clusters for increased reliability
# though passing a zone in the location parameter is also valid
CLUSTER_REGION = "us-west1"
CLUSTER_NAME = "example-cluster"
kubernetes_full_pod = GKEStartPodOperator(
    task_id="ex-all-configs",
    name="full",
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
    cluster_name=CLUSTER_NAME,
    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"},
    ),
    # 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={},
)

Elimina il cluster

Il codice mostrato qui elimina il cluster creato all'inizio della guida.

delete_cluster = GKEDeleteClusterOperator(
    task_id="delete_cluster",
    name=CLUSTER_NAME,
    project_id=PROJECT_ID,
    location=CLUSTER_REGION,
)

Passaggi successivi

• Utilizzare il KubernetesPodOperator
• Gestire i DAG