Source code for airflow.providers.google.cloud.operators.dataproc

#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.
"""This module contains Google Dataproc operators."""

from __future__ import annotations

import inspect
import re
import time
import warnings
from collections.abc import MutableSequence, Sequence
from dataclasses import dataclass
from datetime import datetime, timedelta
from enum import Enum
from functools import cached_property
from typing import TYPE_CHECKING, Any

from google.api_core.exceptions import AlreadyExists, NotFound
from google.api_core.gapic_v1.method import DEFAULT, _MethodDefault
from google.api_core.retry import Retry, exponential_sleep_generator
from google.cloud.dataproc_v1 import Batch, Cluster, ClusterStatus, JobStatus

from airflow.configuration import conf
from airflow.exceptions import AirflowException, AirflowProviderDeprecationWarning
from airflow.providers.google.cloud.hooks.dataproc import (
    DataprocHook,
    DataProcJobBuilder,
    DataprocResourceIsNotReadyError,
)
from airflow.providers.google.cloud.links.dataproc import (
    DATAPROC_BATCH_LINK,
    DATAPROC_CLUSTER_LINK_DEPRECATED,
    DATAPROC_JOB_LINK_DEPRECATED,
    DataprocBatchesListLink,
    DataprocBatchLink,
    DataprocClusterLink,
    DataprocJobLink,
    DataprocLink,
    DataprocWorkflowLink,
    DataprocWorkflowTemplateLink,
)
from airflow.providers.google.cloud.openlineage.utils import (
    inject_openlineage_properties_into_dataproc_job,
)
from airflow.providers.google.cloud.operators.cloud_base import GoogleCloudBaseOperator
from airflow.providers.google.cloud.triggers.dataproc import (
    DataprocBatchTrigger,
    DataprocClusterTrigger,
    DataprocDeleteClusterTrigger,
    DataprocOperationTrigger,
    DataprocSubmitTrigger,
)
from airflow.providers.google.cloud.utils.dataproc import DataprocOperationType
from airflow.providers.google.common.deprecated import deprecated
from airflow.providers.google.common.hooks.base_google import PROVIDE_PROJECT_ID
from airflow.utils import timezone

if TYPE_CHECKING:
    from google.api_core import operation
    from google.api_core.retry_async import AsyncRetry
    from google.protobuf.duration_pb2 import Duration
    from google.protobuf.field_mask_pb2 import FieldMask
    from google.type.interval_pb2 import Interval

    from airflow.utils.context import Context


[docs]class PreemptibilityType(Enum): """Contains possible Type values of Preemptibility applicable for every secondary worker of Cluster."""
[docs] PREEMPTIBLE = "PREEMPTIBLE"
[docs] SPOT = "SPOT"
[docs] PREEMPTIBILITY_UNSPECIFIED = "PREEMPTIBILITY_UNSPECIFIED"
[docs] NON_PREEMPTIBLE = "NON_PREEMPTIBLE"
@dataclass
[docs]class InstanceSelection: """ Defines machines types and a rank to which the machines types belong. Representation for google.cloud.dataproc.v1#google.cloud.dataproc.v1.InstanceFlexibilityPolicy.InstanceSelection. :param machine_types: Full machine-type names, e.g. "n1-standard-16". :param rank: Preference of this instance selection. Lower number means higher preference. Dataproc will first try to create a VM based on the machine-type with priority rank and fallback to next rank based on availability. Machine types and instance selections with the same priority have the same preference. """
[docs] machine_types: list[str]
[docs] rank: int = 0
@dataclass
[docs]class InstanceFlexibilityPolicy: """ Instance flexibility Policy allowing a mixture of VM shapes and provisioning models. Representation for google.cloud.dataproc.v1#google.cloud.dataproc.v1.InstanceFlexibilityPolicy. :param instance_selection_list: List of instance selection options that the group will use when creating new VMs. """
[docs] instance_selection_list: list[InstanceSelection]
[docs]class ClusterGenerator: """ Create a new Dataproc Cluster. :param cluster_name: The name of the DataProc cluster to create. (templated) :param project_id: The ID of the google cloud project in which to create the cluster. (templated) :param num_workers: The # of workers to spin up. If set to zero will spin up cluster in a single node mode :param min_num_workers: The minimum number of primary worker instances to create. If more than ``min_num_workers`` VMs are created out of ``num_workers``, the failed VMs will be deleted, cluster is resized to available VMs and set to RUNNING. If created VMs are less than ``min_num_workers``, the cluster is placed in ERROR state. The failed VMs are not deleted. :param storage_bucket: The storage bucket to use, setting to None lets dataproc generate a custom one for you :param init_actions_uris: List of GCS uri's containing dataproc initialization scripts :param init_action_timeout: Amount of time executable scripts in init_actions_uris has to complete :param metadata: dict of key-value google compute engine metadata entries to add to all instances :param image_version: the version of software inside the Dataproc cluster :param custom_image: custom Dataproc image for more info see https://cloud.google.com/dataproc/docs/guides/dataproc-images :param custom_image_project_id: project id for the custom Dataproc image, for more info see https://cloud.google.com/dataproc/docs/guides/dataproc-images :param custom_image_family: family for the custom Dataproc image, family name can be provide using --family flag while creating custom image, for more info see https://cloud.google.com/dataproc/docs/guides/dataproc-images :param autoscaling_policy: The autoscaling policy used by the cluster. Only resource names including projectid and location (region) are valid. Example: ``projects/[projectId]/locations/[dataproc_region]/autoscalingPolicies/[policy_id]`` :param properties: dict of properties to set on config files (e.g. spark-defaults.conf), see https://cloud.google.com/dataproc/docs/reference/rest/v1/projects.regions.clusters#SoftwareConfig :param optional_components: List of optional cluster components, for more info see https://cloud.google.com/dataproc/docs/reference/rest/v1/ClusterConfig#Component :param num_masters: The # of master nodes to spin up :param master_machine_type: Compute engine machine type to use for the primary node :param master_disk_type: Type of the boot disk for the primary node (default is ``pd-standard``). Valid values: ``pd-ssd`` (Persistent Disk Solid State Drive) or ``pd-standard`` (Persistent Disk Hard Disk Drive). :param master_disk_size: Disk size for the primary node :param master_accelerator_type: Type of the accelerator card (GPU) to attach to the primary node, see https://cloud.google.com/dataproc/docs/reference/rest/v1/InstanceGroupConfig#acceleratorconfig :param master_accelerator_count: Number of accelerator cards (GPUs) to attach to the primary node :param worker_machine_type: Compute engine machine type to use for the worker nodes :param worker_disk_type: Type of the boot disk for the worker node (default is ``pd-standard``). Valid values: ``pd-ssd`` (Persistent Disk Solid State Drive) or ``pd-standard`` (Persistent Disk Hard Disk Drive). :param worker_disk_size: Disk size for the worker nodes :param worker_accelerator_type: Type of the accelerator card (GPU) to attach to the worker nodes, see https://cloud.google.com/dataproc/docs/reference/rest/v1/InstanceGroupConfig#acceleratorconfig :param worker_accelerator_count: Number of accelerator cards (GPUs) to attach to the worker nodes :param num_preemptible_workers: The # of VM instances in the instance group as secondary workers inside the cluster with Preemptibility enabled by default. Note, that it is not possible to mix non-preemptible and preemptible secondary workers in one cluster. :param preemptibility: The type of Preemptibility applicable for every secondary worker, see https://cloud.google.com/dataproc/docs/reference/rpc/ \ google.cloud.dataproc.v1#google.cloud.dataproc.v1.InstanceGroupConfig.Preemptibility :param zone: The zone where the cluster will be located. Set to None to auto-zone. (templated) :param network_uri: The network uri to be used for machine communication, cannot be specified with subnetwork_uri :param subnetwork_uri: The subnetwork uri to be used for machine communication, cannot be specified with network_uri :param internal_ip_only: If true, all instances in the cluster will only have internal IP addresses. This can only be enabled for subnetwork enabled networks :param tags: The GCE tags to add to all instances :param region: The specified region where the dataproc cluster is created. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param service_account: The service account of the dataproc instances. :param service_account_scopes: The URIs of service account scopes to be included. :param idle_delete_ttl: The longest duration that cluster would keep alive while staying idle. Passing this threshold will cause cluster to be auto-deleted. A duration in seconds. :param auto_delete_time: The time when cluster will be auto-deleted. :param auto_delete_ttl: The life duration of cluster, the cluster will be auto-deleted at the end of this duration. A duration in seconds. (If auto_delete_time is set this parameter will be ignored) :param customer_managed_key: The customer-managed key used for disk encryption ``projects/[PROJECT_STORING_KEYS]/locations/[LOCATION]/keyRings/[KEY_RING_NAME]/cryptoKeys/[KEY_NAME]`` # noqa :param enable_component_gateway: Provides access to the web interfaces of default and selected optional components on the cluster. :param driver_pool_size: The number of driver nodes in the node group. :param driver_pool_id: The ID for the driver pool. Must be unique within the cluster. Use this ID to identify the driver group in future operations, such as resizing the node group. :param secondary_worker_instance_flexibility_policy: Instance flexibility Policy allowing a mixture of VM shapes and provisioning models. :param secondary_worker_accelerator_type: Type of the accelerator card (GPU) to attach to the secondary workers, see https://cloud.google.com/dataproc/docs/reference/rest/v1/InstanceGroupConfig#acceleratorconfig :param secondary_worker_accelerator_count: Number of accelerator cards (GPUs) to attach to the secondary workers """ def __init__( self, project_id: str, num_workers: int | None = None, min_num_workers: int | None = None, zone: str | None = None, network_uri: str | None = None, subnetwork_uri: str | None = None, internal_ip_only: bool | None = None, tags: list[str] | None = None, storage_bucket: str | None = None, init_actions_uris: list[str] | None = None, init_action_timeout: str = "10m", metadata: dict | None = None, custom_image: str | None = None, custom_image_project_id: str | None = None, custom_image_family: str | None = None, image_version: str | None = None, autoscaling_policy: str | None = None, properties: dict | None = None, optional_components: list[str] | None = None, num_masters: int = 1, master_machine_type: str = "n1-standard-4", master_disk_type: str = "pd-standard", master_disk_size: int = 1024, master_accelerator_type: str | None = None, master_accelerator_count: int | None = None, worker_machine_type: str = "n1-standard-4", worker_disk_type: str = "pd-standard", worker_disk_size: int = 1024, worker_accelerator_type: str | None = None, worker_accelerator_count: int | None = None, num_preemptible_workers: int = 0, preemptibility: str = PreemptibilityType.PREEMPTIBLE.value, service_account: str | None = None, service_account_scopes: list[str] | None = None, idle_delete_ttl: int | None = None, auto_delete_time: datetime | None = None, auto_delete_ttl: int | None = None, customer_managed_key: str | None = None, enable_component_gateway: bool | None = False, driver_pool_size: int = 0, driver_pool_id: str | None = None, secondary_worker_instance_flexibility_policy: InstanceFlexibilityPolicy | None = None, secondary_worker_accelerator_type: str | None = None, secondary_worker_accelerator_count: int | None = None, **kwargs, ) -> None: self.project_id = project_id self.num_masters = num_masters self.num_workers = num_workers self.min_num_workers = min_num_workers self.num_preemptible_workers = num_preemptible_workers self.preemptibility = self._set_preemptibility_type(preemptibility) self.storage_bucket = storage_bucket self.init_actions_uris = init_actions_uris self.init_action_timeout = init_action_timeout self.metadata = metadata self.custom_image = custom_image self.custom_image_project_id = custom_image_project_id self.custom_image_family = custom_image_family self.image_version = image_version self.properties = properties or {} self.optional_components = optional_components self.master_machine_type = master_machine_type self.master_disk_type = master_disk_type self.master_disk_size = master_disk_size self.master_accelerator_type = master_accelerator_type self.master_accelerator_count = master_accelerator_count self.autoscaling_policy = autoscaling_policy self.worker_machine_type = worker_machine_type self.worker_disk_type = worker_disk_type self.worker_disk_size = worker_disk_size self.worker_accelerator_type = worker_accelerator_type self.worker_accelerator_count = worker_accelerator_count self.zone = zone self.network_uri = network_uri self.subnetwork_uri = subnetwork_uri self.internal_ip_only = internal_ip_only self.tags = tags self.service_account = service_account self.service_account_scopes = service_account_scopes self.idle_delete_ttl = idle_delete_ttl self.auto_delete_time = auto_delete_time self.auto_delete_ttl = auto_delete_ttl self.customer_managed_key = customer_managed_key self.enable_component_gateway = enable_component_gateway self.single_node = num_workers == 0 self.driver_pool_size = driver_pool_size self.driver_pool_id = driver_pool_id self.secondary_worker_instance_flexibility_policy = secondary_worker_instance_flexibility_policy self.secondary_worker_accelerator_type = secondary_worker_accelerator_type self.secondary_worker_accelerator_count = secondary_worker_accelerator_count if self.custom_image and self.image_version: raise ValueError("The custom_image and image_version can't be both set") if self.custom_image_family and self.image_version: raise ValueError("The image_version and custom_image_family can't be both set") if self.custom_image_family and self.custom_image: raise ValueError("The custom_image and custom_image_family can't be both set") if self.single_node and self.num_preemptible_workers > 0: raise ValueError("Single node cannot have preemptible workers.") if self.min_num_workers: if not self.num_workers: raise ValueError("Must specify num_workers when min_num_workers are provided.") if self.min_num_workers > self.num_workers: raise ValueError( "The value of min_num_workers must be less than or equal to num_workers. " f"Provided {self.min_num_workers}(min_num_workers) and {self.num_workers}(num_workers)." ) def _set_preemptibility_type(self, preemptibility: str): return PreemptibilityType(preemptibility.upper()) def _get_init_action_timeout(self) -> dict: match = re.fullmatch(r"(\d+)([sm])", self.init_action_timeout) if match: val = int(match.group(1)) unit = match.group(2) if unit == "s": return {"seconds": val} elif unit == "m": return {"seconds": int(timedelta(minutes=val).total_seconds())} raise AirflowException( "DataprocClusterCreateOperator init_action_timeout" " should be expressed in minutes or seconds. i.e. 10m, 30s" ) def _build_gce_cluster_config(self, cluster_data): # This variable is created since same string was being used multiple times config = "gce_cluster_config" if self.zone: zone_uri = f"https://www.googleapis.com/compute/v1/projects/{self.project_id}/zones/{self.zone}" cluster_data[config]["zone_uri"] = zone_uri if self.metadata: cluster_data[config]["metadata"] = self.metadata if self.network_uri: cluster_data[config]["network_uri"] = self.network_uri if self.subnetwork_uri: cluster_data[config]["subnetwork_uri"] = self.subnetwork_uri if self.internal_ip_only is not None: if not self.subnetwork_uri and self.internal_ip_only: raise AirflowException("Set internal_ip_only to true only when you pass a subnetwork_uri.") cluster_data[config]["internal_ip_only"] = self.internal_ip_only if self.tags: cluster_data[config]["tags"] = self.tags if self.service_account: cluster_data[config]["service_account"] = self.service_account if self.service_account_scopes: cluster_data[config]["service_account_scopes"] = self.service_account_scopes return cluster_data def _build_lifecycle_config(self, cluster_data): # This variable is created since same string was being used multiple times lifecycle_config = "lifecycle_config" if self.idle_delete_ttl: cluster_data[lifecycle_config]["idle_delete_ttl"] = {"seconds": self.idle_delete_ttl} if self.auto_delete_time: utc_auto_delete_time = timezone.convert_to_utc(self.auto_delete_time) cluster_data[lifecycle_config]["auto_delete_time"] = utc_auto_delete_time.strftime( "%Y-%m-%dT%H:%M:%S.%fZ" ) elif self.auto_delete_ttl: cluster_data[lifecycle_config]["auto_delete_ttl"] = {"seconds": int(self.auto_delete_ttl)} return cluster_data def _build_driver_pool(self): driver_pool = { "node_group": { "roles": ["DRIVER"], "node_group_config": {"num_instances": self.driver_pool_size}, }, } if self.driver_pool_id: driver_pool["node_group_id"] = self.driver_pool_id return driver_pool def _build_cluster_data(self): if self.zone: master_type_uri = ( f"projects/{self.project_id}/zones/{self.zone}/machineTypes/{self.master_machine_type}" ) worker_type_uri = ( f"projects/{self.project_id}/zones/{self.zone}/machineTypes/{self.worker_machine_type}" ) else: master_type_uri = self.master_machine_type worker_type_uri = self.worker_machine_type cluster_data = { "gce_cluster_config": {}, "master_config": { "num_instances": self.num_masters, "machine_type_uri": master_type_uri, "disk_config": { "boot_disk_type": self.master_disk_type, "boot_disk_size_gb": self.master_disk_size, }, }, "worker_config": { "num_instances": self.num_workers, "machine_type_uri": worker_type_uri, "disk_config": { "boot_disk_type": self.worker_disk_type, "boot_disk_size_gb": self.worker_disk_size, }, }, "secondary_worker_config": {}, "software_config": {}, "lifecycle_config": {}, "encryption_config": {}, "autoscaling_config": {}, "endpoint_config": {}, } if self.min_num_workers: cluster_data["worker_config"]["min_num_instances"] = self.min_num_workers if self.master_accelerator_type: cluster_data["master_config"]["accelerators"] = { "accelerator_type_uri": self.master_accelerator_type, "accelerator_count": self.master_accelerator_count, } if self.worker_accelerator_type: cluster_data["worker_config"]["accelerators"] = { "accelerator_type_uri": self.worker_accelerator_type, "accelerator_count": self.worker_accelerator_count, } if self.num_preemptible_workers > 0: cluster_data["secondary_worker_config"] = { "num_instances": self.num_preemptible_workers, "machine_type_uri": worker_type_uri, "disk_config": { "boot_disk_type": self.worker_disk_type, "boot_disk_size_gb": self.worker_disk_size, }, "is_preemptible": True, "preemptibility": self.preemptibility.value, } if self.worker_accelerator_type: cluster_data["secondary_worker_config"]["accelerators"] = { "accelerator_type_uri": self.secondary_worker_accelerator_type, "accelerator_count": self.secondary_worker_accelerator_count, } if self.secondary_worker_instance_flexibility_policy: cluster_data["secondary_worker_config"]["instance_flexibility_policy"] = { "instance_selection_list": [ vars(s) for s in self.secondary_worker_instance_flexibility_policy.instance_selection_list ] } if self.storage_bucket: cluster_data["config_bucket"] = self.storage_bucket if self.image_version: cluster_data["software_config"]["image_version"] = self.image_version elif self.custom_image: project_id = self.custom_image_project_id or self.project_id custom_image_url = ( f"https://www.googleapis.com/compute/beta/projects/{project_id}" f"/global/images/{self.custom_image}" ) cluster_data["master_config"]["image_uri"] = custom_image_url if not self.single_node: cluster_data["worker_config"]["image_uri"] = custom_image_url elif self.custom_image_family: project_id = self.custom_image_project_id or self.project_id custom_image_url = ( "https://www.googleapis.com/compute/beta/projects/" f"{project_id}/global/images/family/{self.custom_image_family}" ) cluster_data["master_config"]["image_uri"] = custom_image_url if not self.single_node: cluster_data["worker_config"]["image_uri"] = custom_image_url if self.driver_pool_size > 0: cluster_data["auxiliary_node_groups"] = [self._build_driver_pool()] cluster_data = self._build_gce_cluster_config(cluster_data) if self.single_node: self.properties["dataproc:dataproc.allow.zero.workers"] = "true" if self.properties: cluster_data["software_config"]["properties"] = self.properties if self.optional_components: cluster_data["software_config"]["optional_components"] = self.optional_components cluster_data = self._build_lifecycle_config(cluster_data) if self.init_actions_uris: init_actions_dict = [ {"executable_file": uri, "execution_timeout": self._get_init_action_timeout()} for uri in self.init_actions_uris ] cluster_data["initialization_actions"] = init_actions_dict if self.customer_managed_key: cluster_data["encryption_config"] = {"gce_pd_kms_key_name": self.customer_managed_key} if self.autoscaling_policy: cluster_data["autoscaling_config"] = {"policy_uri": self.autoscaling_policy} if self.enable_component_gateway: cluster_data["endpoint_config"] = {"enable_http_port_access": self.enable_component_gateway} return cluster_data
[docs] def make(self): """ Act as a helper method for easier migration. :return: Dict representing Dataproc cluster. """ return self._build_cluster_data()
[docs]class DataprocCreateClusterOperator(GoogleCloudBaseOperator): """ Create a new cluster on Google Cloud Dataproc. The operator will wait until the creation is successful or an error occurs in the creation process. If the cluster already exists and ``use_if_exists`` is True, then the operator will: - if cluster state is ERROR then delete it if specified and raise error - if cluster state is CREATING wait for it and then check for ERROR state - if cluster state is DELETING wait for it and then create new cluster Please refer to https://cloud.google.com/dataproc/docs/reference/rest/v1/projects.regions.clusters for a detailed explanation on the different parameters. Most of the configuration parameters detailed in the link are available as a parameter to this operator. .. seealso:: For more information on how to use this operator, take a look at the guide: :ref:`howto/operator:DataprocCreateClusterOperator` :param project_id: The ID of the Google cloud project in which to create the cluster. (templated) :param cluster_name: Name of the cluster to create :param labels: Labels that will be assigned to created cluster. Please, notice that adding labels to ClusterConfig object in cluster_config parameter will not lead to adding labels to the cluster. Labels for the clusters could be only set by passing values to parameter of DataprocCreateCluster operator. :param cluster_config: Required. The cluster config to create. If a dict is provided, it must be of the same form as the protobuf message :class:`~google.cloud.dataproc_v1.types.ClusterConfig` :param virtual_cluster_config: Optional. The virtual cluster config, used when creating a Dataproc cluster that does not directly control the underlying compute resources, for example, when creating a `Dataproc-on-GKE cluster <https://cloud.google.com/dataproc/docs/concepts/jobs/dataproc-gke#create-a-dataproc-on-gke-cluster>` :param region: The specified region where the dataproc cluster is created. :param delete_on_error: If true the cluster will be deleted if created with ERROR state. Default value is true. :param use_if_exists: If true use existing cluster :param num_retries_if_resource_is_not_ready: Optional. The number of retry for cluster creation request when resource is not ready error appears. :param request_id: Optional. A unique id used to identify the request. If the server receives two ``DeleteClusterRequest`` requests with the same id, then the second request will be ignored and the first ``google.longrunning.Operation`` created and stored in the backend is returned. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the run status. """
[docs] template_fields: Sequence[str] = ( "project_id", "region", "cluster_config", "virtual_cluster_config", "cluster_name", "labels", "impersonation_chain", )
[docs] template_fields_renderers = {"cluster_config": "json", "virtual_cluster_config": "json"}
def __init__( self, *, cluster_name: str, region: str, project_id: str = PROVIDE_PROJECT_ID, cluster_config: dict | Cluster | None = None, virtual_cluster_config: dict | None = None, labels: dict | None = None, request_id: str | None = None, delete_on_error: bool = True, use_if_exists: bool = True, num_retries_if_resource_is_not_ready: int = 0, retry: AsyncRetry | _MethodDefault | Retry = DEFAULT, timeout: float = 1 * 60 * 60, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, **kwargs, ) -> None: # TODO: remove one day if cluster_config is None and virtual_cluster_config is None: warnings.warn( f"Passing cluster parameters by keywords to `{type(self).__name__}` will be deprecated. " "Please provide cluster_config object using `cluster_config` parameter. " "You can use `airflow.dataproc.ClusterGenerator.generate_cluster` " "method to obtain cluster object.", AirflowProviderDeprecationWarning, stacklevel=2, ) # Remove result of apply defaults if "params" in kwargs: del kwargs["params"] # Create cluster object from kwargs if project_id is None: raise AirflowException( "project_id argument is required when building cluster from keywords parameters" ) kwargs["project_id"] = project_id cluster_config = ClusterGenerator(**kwargs).make() # Remove from kwargs cluster params passed for backward compatibility cluster_params = inspect.signature(ClusterGenerator.__init__).parameters for arg in cluster_params: if arg in kwargs: del kwargs[arg] super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.cluster_config = cluster_config self.cluster_name = cluster_name self.labels = labels self.project_id = project_id self.region = region self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.delete_on_error = delete_on_error self.use_if_exists = use_if_exists self.impersonation_chain = impersonation_chain self.virtual_cluster_config = virtual_cluster_config self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds self.num_retries_if_resource_is_not_ready = num_retries_if_resource_is_not_ready def _create_cluster(self, hook: DataprocHook): return hook.create_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, labels=self.labels, cluster_config=self.cluster_config, virtual_cluster_config=self.virtual_cluster_config, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) def _delete_cluster(self, hook): self.log.info("Deleting the cluster") hook.delete_cluster(region=self.region, cluster_name=self.cluster_name, project_id=self.project_id) def _get_cluster(self, hook: DataprocHook) -> Cluster: return hook.get_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) def _handle_error_state(self, hook: DataprocHook, cluster: Cluster) -> None: if cluster.status.state != cluster.status.State.ERROR: return self.log.info("Cluster is in ERROR state") self.log.info("Gathering diagnostic information.") try: operation = hook.diagnose_cluster( region=self.region, cluster_name=self.cluster_name, project_id=self.project_id ) operation.result() gcs_uri = str(operation.operation.response.value) self.log.info( "Diagnostic information for cluster %s available at: %s", self.cluster_name, gcs_uri ) except Exception as diagnose_error: self.log.info("Some error occurred when trying to diagnose cluster.") self.log.exception(diagnose_error) finally: if self.delete_on_error: self._delete_cluster(hook) # The delete op is asynchronous and can cause further failure if the cluster finishes # deleting between catching AlreadyExists and checking state self._wait_for_cluster_in_deleting_state(hook) raise AirflowException("Cluster was created in an ERROR state then deleted.") raise AirflowException("Cluster was created but is in ERROR state") def _wait_for_cluster_in_deleting_state(self, hook: DataprocHook) -> None: time_left = self.timeout for time_to_sleep in exponential_sleep_generator(initial=10, maximum=120): if time_left < 0: raise AirflowException(f"Cluster {self.cluster_name} is still DELETING state, aborting") time.sleep(time_to_sleep) time_left -= time_to_sleep try: self._get_cluster(hook) except NotFound: break def _wait_for_cluster_in_creating_state(self, hook: DataprocHook) -> Cluster: time_left = self.timeout cluster = self._get_cluster(hook) for time_to_sleep in exponential_sleep_generator(initial=10, maximum=120): if cluster.status.state != cluster.status.State.CREATING: break if time_left < 0: raise AirflowException(f"Cluster {self.cluster_name} is still CREATING state, aborting") time.sleep(time_to_sleep) time_left -= time_to_sleep cluster = self._get_cluster(hook) return cluster def _start_cluster(self, hook: DataprocHook): op: operation.Operation = hook.start_cluster( region=self.region, project_id=self.project_id, cluster_name=self.cluster_name, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) return hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=op) def _retry_cluster_creation(self, hook: DataprocHook): self.log.info("Retrying creation process for Cluster %s", self.cluster_name) self._delete_cluster(hook) self._wait_for_cluster_in_deleting_state(hook) self.log.info("Starting a new creation for Cluster %s", self.cluster_name) operation = self._create_cluster(hook) cluster = hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=operation) self.log.info("Cluster created.") return Cluster.to_dict(cluster)
[docs] def execute(self, context: Context) -> dict: self.log.info("Creating cluster: %s", self.cluster_name) hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) # Save data required to display extra link no matter what the cluster status will be project_id = self.project_id or hook.project_id if project_id: DataprocClusterLink.persist( context=context, operator=self, cluster_id=self.cluster_name, project_id=project_id, region=self.region, ) try: # First try to create a new cluster operation = self._create_cluster(hook) if not self.deferrable and type(operation) is not str: cluster = hook.wait_for_operation( timeout=self.timeout, result_retry=self.retry, operation=operation ) self.log.info("Cluster created.") return Cluster.to_dict(cluster) else: cluster = hook.get_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name ) if cluster.status.state == cluster.status.State.RUNNING: self.log.info("Cluster created.") return Cluster.to_dict(cluster) else: self.defer( trigger=DataprocClusterTrigger( cluster_name=self.cluster_name, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, delete_on_error=self.delete_on_error, ), method_name="execute_complete", ) except AlreadyExists: if not self.use_if_exists: raise self.log.info("Cluster already exists.") cluster = self._get_cluster(hook) except DataprocResourceIsNotReadyError as resource_not_ready_error: if self.num_retries_if_resource_is_not_ready: attempt = self.num_retries_if_resource_is_not_ready while attempt > 0: attempt -= 1 try: cluster = self._retry_cluster_creation(hook) except DataprocResourceIsNotReadyError: continue else: return cluster self.log.info( "Retrying Cluster %s creation because of resource not ready was unsuccessful.", self.cluster_name, ) if self.delete_on_error: self._delete_cluster(hook) self._wait_for_cluster_in_deleting_state(hook) raise resource_not_ready_error except AirflowException as ae: # There still could be a cluster created here in an ERROR state which # should be deleted immediately rather than consuming another retry attempt # (assuming delete_on_error is true (default)) # This reduces overall the number of task attempts from 3 to 2 to successful cluster creation # assuming the underlying GCE issues have resolved within that window. Users can configure # a higher number of retry attempts in powers of two with 30s-60s wait interval try: cluster = self._get_cluster(hook) self._handle_error_state(hook, cluster) except AirflowException as ae_inner: # We could get any number of failures here, including cluster not found and we # can just ignore to ensure we surface the original cluster create failure self.log.exception(ae_inner) finally: raise ae # Check if cluster is not in ERROR state self._handle_error_state(hook, cluster) if cluster.status.state == cluster.status.State.CREATING: # Wait for cluster to be created cluster = self._wait_for_cluster_in_creating_state(hook) self._handle_error_state(hook, cluster) elif cluster.status.state == cluster.status.State.DELETING: # Wait for cluster to be deleted self._wait_for_cluster_in_deleting_state(hook) # Create new cluster cluster = self._create_cluster(hook) self._handle_error_state(hook, cluster) elif cluster.status.state == cluster.status.State.STOPPED: # if the cluster exists and already stopped, then start the cluster self._start_cluster(hook) return Cluster.to_dict(cluster)
[docs] def execute_complete(self, context: Context, event: dict[str, Any]) -> Any: """ Act as a callback for when the trigger fires - returns immediately. Relies on trigger to throw an exception, otherwise it assumes execution was successful. """ cluster_state = event["cluster_state"] cluster_name = event["cluster_name"] if cluster_state == ClusterStatus.State.ERROR: raise AirflowException(f"Cluster is in ERROR state:\n{cluster_name}") self.log.info("%s completed successfully.", self.task_id) return event["cluster"]
# TODO: Remove one day @deprecated( planned_removal_date="March 01, 2025", use_instead="DataprocUpdateClusterOperator", category=AirflowProviderDeprecationWarning, )
[docs]class DataprocScaleClusterOperator(GoogleCloudBaseOperator): """ Scale, up or down, a cluster on Google Cloud Dataproc. The operator will wait until the cluster is re-scaled. Example usage: .. code-block:: python t1 = DataprocClusterScaleOperator( task_id="dataproc_scale", project_id="my-project", cluster_name="cluster-1", num_workers=10, num_preemptible_workers=10, graceful_decommission_timeout="1h", ) .. seealso:: For more detail on about scaling clusters have a look at the reference: https://cloud.google.com/dataproc/docs/concepts/configuring-clusters/scaling-clusters :param cluster_name: The name of the cluster to scale. (templated) :param project_id: The ID of the google cloud project in which the cluster runs. (templated) :param region: The region for the dataproc cluster. (templated) :param num_workers: The new number of workers :param num_preemptible_workers: The new number of preemptible workers :param graceful_decommission_timeout: Timeout for graceful YARN decommissioning. Maximum value is 1d :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). """
[docs] template_fields: Sequence[str] = ("cluster_name", "project_id", "region", "impersonation_chain")
def __init__( self, *, cluster_name: str, project_id: str = PROVIDE_PROJECT_ID, region: str = "global", num_workers: int = 2, num_preemptible_workers: int = 0, graceful_decommission_timeout: str | None = None, gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ) -> None: super().__init__(**kwargs) self.project_id = project_id self.region = region self.cluster_name = cluster_name self.num_workers = num_workers self.num_preemptible_workers = num_preemptible_workers self.graceful_decommission_timeout = graceful_decommission_timeout self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain def _build_scale_cluster_data(self) -> dict: scale_data = { "config": { "worker_config": {"num_instances": self.num_workers}, "secondary_worker_config": {"num_instances": self.num_preemptible_workers}, } } return scale_data @property def _graceful_decommission_timeout_object(self) -> dict[str, int] | None: if not self.graceful_decommission_timeout: return None timeout = None match = re.fullmatch(r"(\d+)([smdh])", self.graceful_decommission_timeout) if match: val = int(match.group(1)) unit = match.group(2) if unit == "s": timeout = val elif unit == "m": timeout = int(timedelta(minutes=val).total_seconds()) elif unit == "h": timeout = int(timedelta(hours=val).total_seconds()) elif unit == "d": timeout = int(timedelta(days=val).total_seconds()) if not timeout: raise AirflowException( "DataprocClusterScaleOperator " " should be expressed in day, hours, minutes or seconds. " " i.e. 1d, 4h, 10m, 30s" ) return {"seconds": timeout}
[docs] def execute(self, context: Context) -> None: """Scale, up or down, a cluster on Google Cloud Dataproc.""" self.log.info("Scaling cluster: %s", self.cluster_name) scaling_cluster_data = self._build_scale_cluster_data() update_mask = ["config.worker_config.num_instances", "config.secondary_worker_config.num_instances"] hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) # Save data required to display extra link no matter what the cluster status will be DataprocLink.persist( context=context, task_instance=self, url=DATAPROC_CLUSTER_LINK_DEPRECATED, resource=self.cluster_name, ) operation = hook.update_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, cluster=scaling_cluster_data, graceful_decommission_timeout=self._graceful_decommission_timeout_object, update_mask={"paths": update_mask}, ) operation.result() self.log.info("Cluster scaling finished")
[docs]class DataprocDeleteClusterOperator(GoogleCloudBaseOperator): """ Delete a cluster in a project. :param region: Required. The Cloud Dataproc region in which to handle the request (templated). :param cluster_name: Required. The cluster name (templated). :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to (templated). :param cluster_uuid: Optional. Specifying the ``cluster_uuid`` means the RPC should fail if cluster with specified UUID does not exist. :param request_id: Optional. A unique id used to identify the request. If the server receives two ``DeleteClusterRequest`` requests with the same id, then the second request will be ignored and the first ``google.longrunning.Operation`` created and stored in the backend is returned. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the cluster status. """
[docs] template_fields: Sequence[str] = ("project_id", "region", "cluster_name", "impersonation_chain")
def __init__( self, *, region: str, cluster_name: str, project_id: str = PROVIDE_PROJECT_ID, cluster_uuid: str | None = None, request_id: str | None = None, retry: AsyncRetry | _MethodDefault = DEFAULT, timeout: float = 1 * 60 * 60, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, **kwargs, ): super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.project_id = project_id self.region = region self.cluster_name = cluster_name self.cluster_uuid = cluster_uuid self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds
[docs] def execute(self, context: Context) -> None: hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) operation = self._delete_cluster(hook) if not self.deferrable: hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=operation) self.log.info("Cluster deleted.") else: try: hook.get_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name ) except NotFound: self.log.info("Cluster deleted.") return except Exception as e: raise AirflowException(str(e)) end_time: float = time.time() + self.timeout self.defer( trigger=DataprocDeleteClusterTrigger( gcp_conn_id=self.gcp_conn_id, project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, end_time=end_time, metadata=self.metadata, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", )
[docs] def execute_complete(self, context: Context, event: dict[str, Any] | None = None) -> Any: """ Act as a callback for when the trigger fires - returns immediately. Relies on trigger to throw an exception, otherwise it assumes execution was successful. """ if event and event["status"] == "error": raise AirflowException(event["message"]) elif event is None: raise AirflowException("No event received in trigger callback") self.log.info("Cluster deleted.")
def _delete_cluster(self, hook: DataprocHook): self.log.info("Deleting cluster: %s", self.cluster_name) return hook.delete_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, cluster_uuid=self.cluster_uuid, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, )
class _DataprocStartStopClusterBaseOperator(GoogleCloudBaseOperator): """ Base class to start or stop a cluster in a project. :param cluster_name: Required. Name of the cluster to create :param region: Required. The specified region where the dataproc cluster is created. :param project_id: Optional. The ID of the Google Cloud project the cluster belongs to. :param cluster_uuid: Optional. Specifying the ``cluster_uuid`` means the RPC should fail if cluster with specified UUID does not exist. :param request_id: Optional. A unique id used to identify the request. If the server receives two ``DeleteClusterRequest`` requests with the same id, then the second request will be ignored and the first ``google.longrunning.Operation`` created and stored in the backend is returned. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). """ template_fields = ( "cluster_name", "region", "project_id", "request_id", "impersonation_chain", ) def __init__( self, *, cluster_name: str, region: str, project_id: str = PROVIDE_PROJECT_ID, cluster_uuid: str | None = None, request_id: str | None = None, retry: AsyncRetry | _MethodDefault | Retry = DEFAULT, timeout: float = 1 * 60 * 60, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ) -> None: super().__init__(**kwargs) self.project_id = project_id self.region = region self.cluster_name = cluster_name self.cluster_uuid = cluster_uuid self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self._hook: DataprocHook | None = None @property def hook(self): if self._hook is None: self._hook = DataprocHook( gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, ) return self._hook def _get_project_id(self) -> str: return self.project_id or self.hook.project_id def _get_cluster(self) -> Cluster: """ Retrieve the cluster information. :return: Instance of ``google.cloud.dataproc_v1.Cluster``` class """ return self.hook.get_cluster( project_id=self._get_project_id(), region=self.region, cluster_name=self.cluster_name, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) def _check_desired_cluster_state(self, cluster: Cluster) -> tuple[bool, str | None]: """ Implement this method in child class to return whether the cluster is in desired state or not. If the cluster is in desired stated you can return a log message content as a second value for the return tuple. :param cluster: Required. Instance of ``google.cloud.dataproc_v1.Cluster`` class to interact with Dataproc API :return: Tuple of (Boolean, Optional[str]) The first value of the tuple is whether the cluster is in desired state or not. The second value of the tuple will use if you want to log something when the cluster is in desired state already. """ raise NotImplementedError def _get_operation(self) -> operation.Operation: """ Implement this method in child class to call the related hook method and return its result. :return: ``google.api_core.operation.Operation`` value whether the cluster is in desired state or not """ raise NotImplementedError def execute(self, context: Context) -> dict | None: cluster: Cluster = self._get_cluster() is_already_desired_state, log_str = self._check_desired_cluster_state(cluster) if is_already_desired_state: self.log.info(log_str) return None op: operation.Operation = self._get_operation() result = self.hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=op) return Cluster.to_dict(result)
[docs]class DataprocStartClusterOperator(_DataprocStartStopClusterBaseOperator): """Start a cluster in a project."""
[docs] def execute(self, context: Context) -> dict | None: self.log.info("Starting the cluster: %s", self.cluster_name) cluster = super().execute(context) DataprocClusterLink.persist( context=context, operator=self, cluster_id=self.cluster_name, project_id=self._get_project_id(), region=self.region, ) self.log.info("Cluster started") return cluster
def _check_desired_cluster_state(self, cluster: Cluster) -> tuple[bool, str | None]: if cluster.status.state == cluster.status.State.RUNNING: return True, f'The cluster "{self.cluster_name}" already running!' return False, None def _get_operation(self) -> operation.Operation: return self.hook.start_cluster( region=self.region, project_id=self._get_project_id(), cluster_name=self.cluster_name, cluster_uuid=self.cluster_uuid, retry=self.retry, timeout=self.timeout, metadata=self.metadata, )
[docs]class DataprocStopClusterOperator(_DataprocStartStopClusterBaseOperator): """Stop a cluster in a project."""
[docs] def execute(self, context: Context) -> dict | None: self.log.info("Stopping the cluster: %s", self.cluster_name) cluster = super().execute(context) self.log.info("Cluster stopped") return cluster
def _check_desired_cluster_state(self, cluster: Cluster) -> tuple[bool, str | None]: if cluster.status.state in [cluster.status.State.STOPPED, cluster.status.State.STOPPING]: return True, f'The cluster "{self.cluster_name}" already stopped!' return False, None def _get_operation(self) -> operation.Operation: return self.hook.stop_cluster( region=self.region, project_id=self._get_project_id(), cluster_name=self.cluster_name, cluster_uuid=self.cluster_uuid, retry=self.retry, timeout=self.timeout, metadata=self.metadata, )
[docs]class DataprocJobBaseOperator(GoogleCloudBaseOperator): """ Base class for operators that launch job on DataProc. :param region: The specified region where the dataproc cluster is created. :param job_name: The job name used in the DataProc cluster. This name by default is the task_id appended with the execution data, but can be templated. The name will always be appended with a random number to avoid name clashes. :param cluster_name: The name of the DataProc cluster. :param project_id: The ID of the Google Cloud project the cluster belongs to, if not specified the project will be inferred from the provided GCP connection. :param dataproc_properties: Map for the Hive properties. Ideal to put in default arguments (templated) :param dataproc_jars: HCFS URIs of jar files to add to the CLASSPATH of the Hive server and Hadoop MapReduce (MR) tasks. Can contain Hive SerDes and UDFs. (templated) :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param labels: The labels to associate with this job. Label keys must contain 1 to 63 characters, and must conform to RFC 1035. Label values may be empty, but, if present, must contain 1 to 63 characters, and must conform to RFC 1035. No more than 32 labels can be associated with a job. :param job_error_states: Job states that should be considered error states. Any states in this set will result in an error being raised and failure of the task. Eg, if the ``CANCELLED`` state should also be considered a task failure, pass in ``{'ERROR', 'CANCELLED'}``. Possible values are currently only ``'ERROR'`` and ``'CANCELLED'``, but could change in the future. Defaults to ``{'ERROR'}``. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param asynchronous: Flag to return after submitting the job to the Dataproc API. This is useful for submitting long running jobs and waiting on them asynchronously using the DataprocJobSensor :param deferrable: Run operator in the deferrable mode :param polling_interval_seconds: time in seconds between polling for job completion. The value is considered only when running in deferrable mode. Must be greater than 0. :var dataproc_job_id: The actual "jobId" as submitted to the Dataproc API. This is useful for identifying or linking to the job in the Google Cloud Console Dataproc UI, as the actual "jobId" submitted to the Dataproc API is appended with an 8 character random string. :vartype dataproc_job_id: str """
[docs] job_type = ""
def __init__( self, *, region: str, job_name: str = "{{task.task_id}}_{{ds_nodash}}", cluster_name: str = "cluster-1", project_id: str = PROVIDE_PROJECT_ID, dataproc_properties: dict | None = None, dataproc_jars: list[str] | None = None, gcp_conn_id: str = "google_cloud_default", labels: dict | None = None, job_error_states: set[str] | None = None, impersonation_chain: str | Sequence[str] | None = None, asynchronous: bool = False, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, **kwargs, ) -> None: super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.gcp_conn_id = gcp_conn_id self.labels = labels self.job_name = job_name self.cluster_name = cluster_name self.dataproc_properties = dataproc_properties self.dataproc_jars = dataproc_jars self.region = region self.job_error_states = job_error_states or {"ERROR"} self.impersonation_chain = impersonation_chain self.hook = DataprocHook(gcp_conn_id=gcp_conn_id, impersonation_chain=impersonation_chain) self.project_id = project_id or self.hook.project_id self.job_template: DataProcJobBuilder | None = None self.job: dict | None = None self.dataproc_job_id = None self.asynchronous = asynchronous self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds
[docs] def create_job_template(self) -> DataProcJobBuilder: """Initialize `self.job_template` with default values.""" if self.project_id is None: raise AirflowException( "project id should either be set via project_id " "parameter or retrieved from the connection," ) job_template = DataProcJobBuilder( project_id=self.project_id, task_id=self.task_id, cluster_name=self.cluster_name, job_type=self.job_type, properties=self.dataproc_properties, ) job_template.set_job_name(self.job_name) job_template.add_jar_file_uris(self.dataproc_jars) job_template.add_labels(self.labels) self.job_template = job_template return job_template
def _generate_job_template(self) -> str: if self.job_template: job = self.job_template.build() return job["job"] raise AirflowException("Create a job template before")
[docs] def execute(self, context: Context): if self.job_template: self.job = self.job_template.build() if self.job is None: raise AirflowException("The job should be set here.") self.dataproc_job_id = self.job["job"]["reference"]["job_id"] self.log.info("Submitting %s job %s", self.job_type, self.dataproc_job_id) job_object = self.hook.submit_job( project_id=self.project_id, job=self.job["job"], region=self.region ) job_id = job_object.reference.job_id self.log.info("Job %s submitted successfully.", job_id) # Save data required for extra links no matter what the job status will be DataprocLink.persist( context=context, task_instance=self, url=DATAPROC_JOB_LINK_DEPRECATED, resource=job_id ) if self.deferrable: self.defer( trigger=DataprocSubmitTrigger( job_id=job_id, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", ) if not self.asynchronous: self.log.info("Waiting for job %s to complete", job_id) self.hook.wait_for_job(job_id=job_id, region=self.region, project_id=self.project_id) self.log.info("Job %s completed successfully.", job_id) return job_id else: raise AirflowException("Create a job template before")
[docs] def execute_complete(self, context, event=None) -> None: """ Act as a callback for when the trigger fires - returns immediately. Relies on trigger to throw an exception, otherwise it assumes execution was successful. """ job_state = event["job_state"] job_id = event["job_id"] if job_state == JobStatus.State.ERROR: raise AirflowException(f"Job failed:\n{job_id}") if job_state == JobStatus.State.CANCELLED: raise AirflowException(f"Job was cancelled:\n{job_id}") self.log.info("%s completed successfully.", self.task_id) return job_id
[docs] def on_kill(self) -> None: """Act as a callback called when the operator is killed; cancel any running job.""" if self.dataproc_job_id: self.hook.cancel_job(project_id=self.project_id, job_id=self.dataproc_job_id, region=self.region)
[docs]class DataprocCreateWorkflowTemplateOperator(GoogleCloudBaseOperator): """ Creates new workflow template. :param project_id: Optional. The ID of the Google Cloud project the cluster belongs to. :param region: Required. The Cloud Dataproc region in which to handle the request. :param template: The Dataproc workflow template to create. If a dict is provided, it must be of the same form as the protobuf message WorkflowTemplate. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. """
[docs] template_fields: Sequence[str] = ("region", "template")
[docs] template_fields_renderers = {"template": "json"}
def __init__( self, *, template: dict, region: str, project_id: str = PROVIDE_PROJECT_ID, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ): super().__init__(**kwargs) self.region = region self.template = template self.project_id = project_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Creating template") try: workflow = hook.create_workflow_template( region=self.region, template=self.template, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) self.log.info("Workflow %s created", workflow.name) except AlreadyExists: self.log.info("Workflow with given id already exists") project_id = self.project_id or hook.project_id if project_id: DataprocWorkflowTemplateLink.persist( context=context, operator=self, workflow_template_id=self.template["id"], region=self.region, project_id=project_id, )
[docs]class DataprocInstantiateWorkflowTemplateOperator(GoogleCloudBaseOperator): """ Instantiate a WorkflowTemplate on Google Cloud Dataproc. The operator will wait until the WorkflowTemplate is finished executing. .. seealso:: Please refer to: https://cloud.google.com/dataproc/docs/reference/rest/v1/projects.regions.workflowTemplates/instantiate :param template_id: The id of the template. (templated) :param project_id: The ID of the google cloud project in which the template runs :param region: The specified region where the dataproc cluster is created. :param parameters: a map of parameters for Dataproc Template in key-value format: map (key: string, value: string) Example: { "date_from": "2019-08-01", "date_to": "2019-08-02"}. Values may not exceed 100 characters. Please refer to: https://cloud.google.com/dataproc/docs/concepts/workflows/workflow-parameters :param request_id: Optional. A unique id used to identify the request. If the server receives two ``SubmitJobRequest`` requests with the same id, then the second request will be ignored and the first ``Job`` created and stored in the backend is returned. It is recommended to always set this value to a UUID. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the run status. :param cancel_on_kill: Flag which indicates whether cancel the workflow, when on_kill is called """
[docs] template_fields: Sequence[str] = ("template_id", "impersonation_chain", "request_id", "parameters")
[docs] template_fields_renderers = {"parameters": "json"}
def __init__( self, *, template_id: str, region: str, project_id: str = PROVIDE_PROJECT_ID, version: int | None = None, request_id: str | None = None, parameters: dict[str, str] | None = None, retry: AsyncRetry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, cancel_on_kill: bool = True, **kwargs, ) -> None: super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.template_id = template_id self.parameters = parameters self.version = version self.project_id = project_id self.region = region self.retry = retry self.timeout = timeout self.metadata = metadata self.request_id = request_id self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds self.cancel_on_kill = cancel_on_kill self.operation_name: str | None = None
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Instantiating template %s", self.template_id) operation = hook.instantiate_workflow_template( project_id=self.project_id, region=self.region, template_name=self.template_id, version=self.version, request_id=self.request_id, parameters=self.parameters, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) operation_name = operation.operation.name self.operation_name = operation_name workflow_id = operation_name.split("/")[-1] project_id = self.project_id or hook.project_id if project_id: DataprocWorkflowLink.persist( context=context, operator=self, workflow_id=workflow_id, region=self.region, project_id=project_id, ) self.log.info("Template instantiated. Workflow Id : %s", workflow_id) if not self.deferrable: hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=operation) self.log.info("Workflow %s completed successfully", workflow_id) else: self.defer( trigger=DataprocOperationTrigger( name=operation_name, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", )
[docs] def execute_complete(self, context, event=None) -> None: """ Act as a callback for when the trigger fires. This returns immediately. It relies on trigger to throw an exception, otherwise it assumes execution was successful. """ if event["status"] in ("failed", "error"): self.log.exception("Unexpected error in the operation.") raise AirflowException(event["message"]) self.log.info("Workflow %s completed successfully", event["operation_name"])
[docs] def on_kill(self) -> None: if self.cancel_on_kill and self.operation_name: hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) hook.get_operations_client(region=self.region).cancel_operation(name=self.operation_name)
[docs]class DataprocInstantiateInlineWorkflowTemplateOperator(GoogleCloudBaseOperator): """ Instantiate a WorkflowTemplate Inline on Google Cloud Dataproc. The operator will wait until the WorkflowTemplate is finished executing. .. seealso:: For more information on how to use this operator, take a look at the guide: :ref:`howto/operator:DataprocInstantiateInlineWorkflowTemplateOperator` For more detail on about instantiate inline have a look at the reference: https://cloud.google.com/dataproc/docs/reference/rest/v1/projects.regions.workflowTemplates/instantiateInline :param template: The template contents. (templated) :param project_id: The ID of the google cloud project in which the template runs :param region: The specified region where the dataproc cluster is created. :param parameters: a map of parameters for Dataproc Template in key-value format: map (key: string, value: string) Example: { "date_from": "2019-08-01", "date_to": "2019-08-02"}. Values may not exceed 100 characters. Please refer to: https://cloud.google.com/dataproc/docs/concepts/workflows/workflow-parameters :param request_id: Optional. A unique id used to identify the request. If the server receives two ``SubmitJobRequest`` requests with the same id, then the second request will be ignored and the first ``Job`` created and stored in the backend is returned. It is recommended to always set this value to a UUID. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the run status. :param cancel_on_kill: Flag which indicates whether cancel the workflow, when on_kill is called """
[docs] template_fields: Sequence[str] = ("template", "impersonation_chain")
[docs] template_fields_renderers = {"template": "json"}
def __init__( self, *, template: dict, region: str, project_id: str = PROVIDE_PROJECT_ID, request_id: str | None = None, retry: AsyncRetry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, cancel_on_kill: bool = True, **kwargs, ) -> None: super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.template = template self.project_id = project_id self.region = region self.template = template self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds self.cancel_on_kill = cancel_on_kill self.operation_name: str | None = None
[docs] def execute(self, context: Context): self.log.info("Instantiating Inline Template") hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) project_id = self.project_id or hook.project_id operation = hook.instantiate_inline_workflow_template( template=self.template, project_id=project_id, region=self.region, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) operation_name = operation.operation.name self.operation_name = operation_name workflow_id = operation_name.split("/")[-1] if project_id: DataprocWorkflowLink.persist( context=context, operator=self, workflow_id=workflow_id, region=self.region, project_id=project_id, ) if not self.deferrable: self.log.info("Template instantiated. Workflow Id : %s", workflow_id) hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=operation) self.log.info("Workflow %s completed successfully", workflow_id) else: self.defer( trigger=DataprocOperationTrigger( name=operation_name, project_id=self.project_id or hook.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", )
[docs] def execute_complete(self, context, event=None) -> None: """ Act as a callback for when the trigger fires. This returns immediately. It relies on trigger to throw an exception, otherwise it assumes execution was successful. """ if event["status"] in ("failed", "error"): self.log.exception("Unexpected error in the operation.") raise AirflowException(event["message"]) self.log.info("Workflow %s completed successfully", event["operation_name"])
[docs] def on_kill(self) -> None: if self.cancel_on_kill and self.operation_name: hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) hook.get_operations_client(region=self.region).cancel_operation(name=self.operation_name)
[docs]class DataprocSubmitJobOperator(GoogleCloudBaseOperator): """ Submit a job to a cluster. :param project_id: Optional. The ID of the Google Cloud project that the job belongs to. :param region: Required. The Cloud Dataproc region in which to handle the request. :param job: Required. The job resource. If a dict is provided, it must be of the same form as the protobuf message :class:`~google.cloud.dataproc_v1.types.Job`. For the complete list of supported job types and their configurations please take a look here https://cloud.google.com/dataproc/docs/reference/rest/v1/projects.regions.jobs :param request_id: Optional. A unique id used to identify the request. If the server receives two ``SubmitJobRequest`` requests with the same id, then the second request will be ignored and the first ``Job`` created and stored in the backend is returned. It is recommended to always set this value to a UUID. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param asynchronous: Flag to return after submitting the job to the Dataproc API. This is useful for submitting long-running jobs and waiting on them asynchronously using the DataprocJobSensor :param deferrable: Run operator in the deferrable mode :param polling_interval_seconds: time in seconds between polling for job completion. The value is considered only when running in deferrable mode. Must be greater than 0. :param cancel_on_kill: Flag which indicates whether cancel the hook's job or not, when on_kill is called :param wait_timeout: How many seconds wait for job to be ready. Used only if ``asynchronous`` is False """
[docs] template_fields: Sequence[str] = ("project_id", "region", "job", "impersonation_chain", "request_id")
[docs] template_fields_renderers = {"job": "json"}
def __init__( self, *, job: dict, region: str, project_id: str = PROVIDE_PROJECT_ID, request_id: str | None = None, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, asynchronous: bool = False, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, cancel_on_kill: bool = True, wait_timeout: int | None = None, openlineage_inject_parent_job_info: bool = conf.getboolean( "openlineage", "spark_inject_parent_job_info", fallback=False ), **kwargs, ) -> None: super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.project_id = project_id self.region = region self.job = job self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.asynchronous = asynchronous self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds self.cancel_on_kill = cancel_on_kill self.hook: DataprocHook | None = None self.job_id: str | None = None self.wait_timeout = wait_timeout self.openlineage_inject_parent_job_info = openlineage_inject_parent_job_info
[docs] def execute(self, context: Context): self.log.info("Submitting job") self.hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) if self.openlineage_inject_parent_job_info: self.log.info("Automatic injection of OpenLineage information into Spark properties is enabled.") self.job = inject_openlineage_properties_into_dataproc_job( job=self.job, context=context, inject_parent_job_info=self.openlineage_inject_parent_job_info ) job_object = self.hook.submit_job( project_id=self.project_id, region=self.region, job=self.job, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) new_job_id: str = job_object.reference.job_id self.log.info("Job %s submitted successfully.", new_job_id) # Save data required by extra links no matter what the job status will be project_id = self.project_id or self.hook.project_id if project_id: DataprocJobLink.persist( context=context, operator=self, job_id=new_job_id, region=self.region, project_id=project_id, ) self.job_id = new_job_id if self.deferrable: job = self.hook.get_job(project_id=self.project_id, region=self.region, job_id=self.job_id) state = job.status.state if state == JobStatus.State.DONE: return self.job_id elif state == JobStatus.State.ERROR: raise AirflowException(f"Job failed:\n{job}") elif state == JobStatus.State.CANCELLED: raise AirflowException(f"Job was cancelled:\n{job}") self.defer( trigger=DataprocSubmitTrigger( job_id=self.job_id, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, cancel_on_kill=self.cancel_on_kill, ), method_name="execute_complete", ) elif not self.asynchronous: self.log.info("Waiting for job %s to complete", new_job_id) self.hook.wait_for_job( job_id=new_job_id, region=self.region, project_id=self.project_id, timeout=self.wait_timeout ) self.log.info("Job %s completed successfully.", new_job_id) return self.job_id
[docs] def execute_complete(self, context, event=None) -> None: """ Act as a callback for when the trigger fires. This returns immediately. It relies on trigger to throw an exception, otherwise it assumes execution was successful. """ job_state = event["job_state"] job_id = event["job_id"] job = event["job"] if job_state == JobStatus.State.ERROR: raise AirflowException(f"Job {job_id} failed:\n{job}") if job_state == JobStatus.State.CANCELLED: raise AirflowException(f"Job {job_id} was cancelled:\n{job}") self.log.info("%s completed successfully.", self.task_id) return job_id
[docs] def on_kill(self): if self.job_id and self.cancel_on_kill: self.hook.cancel_job(job_id=self.job_id, project_id=self.project_id, region=self.region)
[docs]class DataprocUpdateClusterOperator(GoogleCloudBaseOperator): """ Update a cluster in a project. :param region: Required. The Cloud Dataproc region in which to handle the request. :param project_id: Optional. The ID of the Google Cloud project the cluster belongs to. :param cluster_name: Required. The cluster name. :param cluster: Required. The changes to the cluster. If a dict is provided, it must be of the same form as the protobuf message :class:`~google.cloud.dataproc_v1.types.Cluster` :param update_mask: Required. Specifies the path, relative to ``Cluster``, of the field to update. For example, to change the number of workers in a cluster to 5, the ``update_mask`` parameter would be specified as ``config.worker_config.num_instances``, and the ``PATCH`` request body would specify the new value. If a dict is provided, it must be of the same form as the protobuf message :class:`~google.protobuf.field_mask_pb2.FieldMask` :param graceful_decommission_timeout: Optional. Timeout for graceful YARN decommissioning. Graceful decommissioning allows removing nodes from the cluster without interrupting jobs in progress. Timeout specifies how long to wait for jobs in progress to finish before forcefully removing nodes (and potentially interrupting jobs). Default timeout is 0 (for forceful decommission), and the maximum allowed timeout is 1 day. :param request_id: Optional. A unique id used to identify the request. If the server receives two ``UpdateClusterRequest`` requests with the same id, then the second request will be ignored and the first ``google.long-running.Operation`` created and stored in the backend is returned. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the run status. """
[docs] template_fields: Sequence[str] = ( "cluster_name", "cluster", "region", "request_id", "project_id", "impersonation_chain", )
def __init__( self, *, cluster_name: str, cluster: dict | Cluster, update_mask: dict | FieldMask, graceful_decommission_timeout: dict | Duration, region: str, request_id: str | None = None, project_id: str = PROVIDE_PROJECT_ID, retry: AsyncRetry | _MethodDefault | Retry = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, **kwargs, ): super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.project_id = project_id self.region = region self.cluster_name = cluster_name self.cluster = cluster self.update_mask = update_mask self.graceful_decommission_timeout = graceful_decommission_timeout self.request_id = request_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) # Save data required by extra links no matter what the cluster status will be project_id = self.project_id or hook.project_id if project_id: DataprocClusterLink.persist( context=context, operator=self, cluster_id=self.cluster_name, project_id=project_id, region=self.region, ) self.log.info("Updating %s cluster.", self.cluster_name) operation = hook.update_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name, cluster=self.cluster, update_mask=self.update_mask, graceful_decommission_timeout=self.graceful_decommission_timeout, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) if not self.deferrable: hook.wait_for_operation(timeout=self.timeout, result_retry=self.retry, operation=operation) else: cluster = hook.get_cluster( project_id=self.project_id, region=self.region, cluster_name=self.cluster_name ) if cluster.status.state != cluster.status.State.RUNNING: self.defer( trigger=DataprocClusterTrigger( cluster_name=self.cluster_name, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", ) self.log.info("Updated %s cluster.", self.cluster_name)
[docs] def execute_complete(self, context: Context, event: dict[str, Any]) -> Any: """ Act as a callback for when the trigger fires - returns immediately. Relies on trigger to throw an exception, otherwise it assumes execution was successful. """ cluster_state = event["cluster_state"] cluster_name = event["cluster_name"] if cluster_state == ClusterStatus.State.ERROR: raise AirflowException(f"Cluster is in ERROR state:\n{cluster_name}") self.log.info("%s completed successfully.", self.task_id)
[docs]class DataprocDiagnoseClusterOperator(GoogleCloudBaseOperator): """ Diagnose a cluster in a project. After the operation completes, the response contains the Cloud Storage URI of the diagnostic output report containing a summary of collected diagnostics. :param region: Required. The Cloud Dataproc region in which to handle the request (templated). :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to (templated). :param cluster_name: Required. The cluster name (templated). :param tarball_gcs_dir: The output Cloud Storage directory for the diagnostic tarball. If not specified, a task-specific directory in the cluster's staging bucket will be used. :param diagnosis_interval: Time interval in which diagnosis should be carried out on the cluster. :param jobs: Specifies a list of jobs on which diagnosis is to be performed. Format: `projects/{project}/regions/{region}/jobs/{job}` :param yarn_application_ids: Specifies a list of yarn applications on which diagnosis is to be performed. :param metadata: Additional metadata that is provided to the method. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the cluster status. """
[docs] template_fields: Sequence[str] = ( "project_id", "region", "cluster_name", "impersonation_chain", "tarball_gcs_dir", "diagnosis_interval", "jobs", "yarn_application_ids", )
def __init__( self, *, region: str, cluster_name: str, project_id: str = PROVIDE_PROJECT_ID, tarball_gcs_dir: str | None = None, diagnosis_interval: dict | Interval | None = None, jobs: MutableSequence[str] | None = None, yarn_application_ids: MutableSequence[str] | None = None, retry: AsyncRetry | _MethodDefault = DEFAULT, timeout: float = 1 * 60 * 60, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 10, **kwargs, ): super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.project_id = project_id self.region = region self.cluster_name = cluster_name self.tarball_gcs_dir = tarball_gcs_dir self.diagnosis_interval = diagnosis_interval self.jobs = jobs self.yarn_application_ids = yarn_application_ids self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Collecting diagnostic tarball for cluster: %s", self.cluster_name) operation = hook.diagnose_cluster( region=self.region, cluster_name=self.cluster_name, project_id=self.project_id, tarball_gcs_dir=self.tarball_gcs_dir, diagnosis_interval=self.diagnosis_interval, jobs=self.jobs, yarn_application_ids=self.yarn_application_ids, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) if not self.deferrable: result = hook.wait_for_operation( timeout=self.timeout, result_retry=self.retry, operation=operation ) self.log.info( "The diagnostic output for cluster %s is available at: %s", self.cluster_name, result.output_uri, ) else: self.defer( trigger=DataprocOperationTrigger( name=operation.operation.name, operation_type=DataprocOperationType.DIAGNOSE.value, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", )
[docs] def execute_complete(self, context: Context, event: dict[str, Any] | None = None) -> None: """ Act as a callback for when the trigger fires. This returns immediately. It relies on trigger to throw an exception, otherwise it assumes execution was successful. """ if event: status = event.get("status") if status in ("failed", "error"): self.log.exception("Unexpected error in the operation.") raise AirflowException(event.get("message")) self.log.info( "The diagnostic output for cluster %s is available at: %s", self.cluster_name, event.get("output_uri"), )
[docs]class DataprocCreateBatchOperator(GoogleCloudBaseOperator): """ Create a batch workload. :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to. (templated) :param region: Required. The Cloud Dataproc region in which to handle the request. (templated) :param batch: Required. The batch to create. (templated) :param batch_id: Required. The ID to use for the batch, which will become the final component of the batch's resource name. This value must be 4-63 characters. Valid characters are /[a-z][0-9]-/. (templated) :param request_id: Optional. A unique id used to identify the request. If the server receives two ``CreateBatchRequest`` requests with the same id, then the second request will be ignored and the first ``google.longrunning.Operation`` created and stored in the backend is returned. :param num_retries_if_resource_is_not_ready: Optional. The number of retry for cluster creation request when resource is not ready error appears. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param result_retry: Result retry object used to retry requests. Is used to decrease delay between executing chained tasks in a DAG by specifying exact amount of seconds for executing. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param asynchronous: Flag to return after creating batch to the Dataproc API. This is useful for creating long-running batch and waiting on them asynchronously using the DataprocBatchSensor :param deferrable: Run operator in the deferrable mode. :param polling_interval_seconds: Time (seconds) to wait between calls to check the run status. """
[docs] template_fields: Sequence[str] = ( "project_id", "batch", "batch_id", "region", "impersonation_chain", )
def __init__( self, *, region: str, project_id: str = PROVIDE_PROJECT_ID, batch: dict | Batch, batch_id: str | None = None, request_id: str | None = None, num_retries_if_resource_is_not_ready: int = 0, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, result_retry: AsyncRetry | _MethodDefault | Retry = DEFAULT, asynchronous: bool = False, deferrable: bool = conf.getboolean("operators", "default_deferrable", fallback=False), polling_interval_seconds: int = 5, **kwargs, ): super().__init__(**kwargs) if deferrable and polling_interval_seconds <= 0: raise ValueError("Invalid value for polling_interval_seconds. Expected value greater than 0") self.region = region self.project_id = project_id self.batch = batch self.batch_id = batch_id self.request_id = request_id self.num_retries_if_resource_is_not_ready = num_retries_if_resource_is_not_ready self.retry = retry self.result_retry = result_retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.operation: operation.Operation | None = None self.asynchronous = asynchronous self.deferrable = deferrable self.polling_interval_seconds = polling_interval_seconds
[docs] def execute(self, context: Context): if self.asynchronous and self.deferrable: raise AirflowException( "Both asynchronous and deferrable parameters were passed. Please, provide only one." ) batch_id: str = "" if self.batch_id: batch_id = self.batch_id self.log.info("Starting batch %s", batch_id) # Persist the link earlier so users can observe the progress DataprocBatchLink.persist( context=context, operator=self, project_id=self.project_id, region=self.region, batch_id=self.batch_id, ) else: self.log.info("Starting batch. The batch ID will be generated since it was not provided.") try: self.operation = self.hook.create_batch( region=self.region, project_id=self.project_id, batch=self.batch, batch_id=self.batch_id, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) except AlreadyExists: self.log.info("Batch with given id already exists.") self.log.info("Attaching to the job %s if it is still running.", batch_id) else: batch_id = self.operation.metadata.batch.split("/")[-1] self.log.info("The batch %s was created.", batch_id) DataprocBatchLink.persist( context=context, operator=self, project_id=self.project_id, region=self.region, batch_id=batch_id, ) if self.asynchronous: batch = self.hook.get_batch( batch_id=batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) self.log.info("The batch %s was created asynchronously. Exiting.", batch_id) return Batch.to_dict(batch) if self.deferrable: self.defer( trigger=DataprocBatchTrigger( batch_id=batch_id, project_id=self.project_id, region=self.region, gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain, polling_interval_seconds=self.polling_interval_seconds, ), method_name="execute_complete", ) self.log.info("Waiting for the completion of batch job %s", batch_id) batch = self.hook.wait_for_batch( batch_id=batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) if self.num_retries_if_resource_is_not_ready and self.hook.check_error_for_resource_is_not_ready_msg( batch.state_message ): attempt = self.num_retries_if_resource_is_not_ready while attempt > 0: attempt -= 1 batch, batch_id = self.retry_batch_creation(batch_id) if not self.hook.check_error_for_resource_is_not_ready_msg(batch.state_message): break self.handle_batch_status(context, batch.state, batch_id, batch.state_message) return Batch.to_dict(batch)
@cached_property
[docs] def hook(self) -> DataprocHook: return DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain)
[docs] def execute_complete(self, context, event=None) -> None: """ Act as a callback for when the trigger fires. This returns immediately. It relies on trigger to throw an exception, otherwise it assumes execution was successful. """ if event is None: raise AirflowException("Batch failed.") state = event["batch_state"] batch_id = event["batch_id"] self.handle_batch_status(context, state, batch_id, state_message=event["batch_state_message"])
[docs] def on_kill(self): if self.operation: self.operation.cancel()
[docs] def handle_batch_status( self, context: Context, state: Batch.State, batch_id: str, state_message: str | None = None ) -> None: # The existing batch may be a number of states other than 'SUCCEEDED'\ # wait_for_operation doesn't fail if the job is cancelled, so we will check for it here which also # finds a cancelling|canceled|unspecified job from wait_for_batch or the deferred trigger link = DATAPROC_BATCH_LINK.format(region=self.region, project_id=self.project_id, batch_id=batch_id) if state == Batch.State.FAILED: raise AirflowException( f"Batch job {batch_id} failed with error: {state_message}\nDriver Logs: {link}" ) if state in (Batch.State.CANCELLED, Batch.State.CANCELLING): raise AirflowException(f"Batch job {batch_id} was cancelled. Driver logs: {link}") if state == Batch.State.STATE_UNSPECIFIED: raise AirflowException(f"Batch job {batch_id} unspecified. Driver logs: {link}") self.log.info("Batch job %s completed. Driver logs: %s", batch_id, link)
[docs] def retry_batch_creation( self, previous_batch_id: str, ): self.log.info("Retrying creation process for batch_id %s", self.batch_id) self.log.info("Deleting previous failed Batch") self.hook.delete_batch( batch_id=previous_batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) self.log.info("Starting a new creation for batch_id %s", self.batch_id) try: self.operation = self.hook.create_batch( region=self.region, project_id=self.project_id, batch=self.batch, batch_id=self.batch_id, request_id=self.request_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) except AlreadyExists: self.log.info("Batch with given id already exists.") self.log.info("Attaching to the job %s if it is still running.", self.batch_id) else: batch_id = self.operation.metadata.batch.split("/")[-1] self.log.info("The batch %s was created.", batch_id) self.log.info("Waiting for the completion of batch job %s", batch_id) batch = self.hook.wait_for_batch( batch_id=batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) return batch, batch_id
[docs]class DataprocDeleteBatchOperator(GoogleCloudBaseOperator): """ Delete the batch workload resource. :param batch_id: Required. The ID to use for the batch, which will become the final component of the batch's resource name. This value must be 4-63 characters. Valid characters are /[a-z][0-9]-/. :param region: Required. The Cloud Dataproc region in which to handle the request. :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). """
[docs] template_fields: Sequence[str] = ("batch_id", "region", "project_id", "impersonation_chain")
def __init__( self, *, batch_id: str, region: str, project_id: str = PROVIDE_PROJECT_ID, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ): super().__init__(**kwargs) self.batch_id = batch_id self.region = region self.project_id = project_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Deleting batch: %s", self.batch_id) hook.delete_batch( batch_id=self.batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) self.log.info("Batch deleted.")
[docs]class DataprocGetBatchOperator(GoogleCloudBaseOperator): """ Get the batch workload resource representation. :param batch_id: Required. The ID to use for the batch, which will become the final component of the batch's resource name. This value must be 4-63 characters. Valid characters are /[a-z][0-9]-/. :param region: Required. The Cloud Dataproc region in which to handle the request. :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). """
[docs] template_fields: Sequence[str] = ("batch_id", "region", "project_id", "impersonation_chain")
def __init__( self, *, batch_id: str, region: str, project_id: str = PROVIDE_PROJECT_ID, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ): super().__init__(**kwargs) self.batch_id = batch_id self.region = region self.project_id = project_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Getting batch: %s", self.batch_id) batch = hook.get_batch( batch_id=self.batch_id, region=self.region, project_id=self.project_id, retry=self.retry, timeout=self.timeout, metadata=self.metadata, ) project_id = self.project_id or hook.project_id if project_id: DataprocBatchLink.persist( context=context, operator=self, project_id=project_id, region=self.region, batch_id=self.batch_id, ) return Batch.to_dict(batch)
[docs]class DataprocListBatchesOperator(GoogleCloudBaseOperator): """ List batch workloads. :param region: Required. The Cloud Dataproc region in which to handle the request. :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to. :param page_size: Optional. The maximum number of batches to return in each response. The service may return fewer than this value. The default page size is 20; the maximum page size is 1000. :param page_token: Optional. A page token received from a previous ``ListBatches`` call. Provide this token to retrieve the subsequent page. :param retry: Optional, a retry object used to retry requests. If `None` is specified, requests will not be retried. :param timeout: Optional, the amount of time, in seconds, to wait for the request to complete. Note that if `retry` is specified, the timeout applies to each individual attempt. :param metadata: Optional, additional metadata that is provided to the method. :param gcp_conn_id: Optional, the connection ID used to connect to Google Cloud Platform. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). :param filter: Result filters as specified in ListBatchesRequest :param order_by: How to order results as specified in ListBatchesRequest """
[docs] template_fields: Sequence[str] = ("region", "project_id", "impersonation_chain")
def __init__( self, *, region: str, project_id: str = PROVIDE_PROJECT_ID, page_size: int | None = None, page_token: str | None = None, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, filter: str | None = None, order_by: str | None = None, **kwargs, ) -> None: super().__init__(**kwargs) self.region = region self.project_id = project_id self.page_size = page_size self.page_token = page_token self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain self.filter = filter self.order_by = order_by
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) results = hook.list_batches( region=self.region, project_id=self.project_id, page_size=self.page_size, page_token=self.page_token, retry=self.retry, timeout=self.timeout, metadata=self.metadata, filter=self.filter, order_by=self.order_by, ) project_id = self.project_id or hook.project_id if project_id: DataprocBatchesListLink.persist(context=context, operator=self, project_id=project_id) return [Batch.to_dict(result) for result in results]
[docs]class DataprocCancelOperationOperator(GoogleCloudBaseOperator): """ Cancel the batch workload resource. :param operation_name: Required. The name of the operation resource to be cancelled. :param region: Required. The Cloud Dataproc region in which to handle the request. :param project_id: Optional. The ID of the Google Cloud project that the cluster belongs to. :param retry: A retry object used to retry requests. If ``None`` is specified, requests will not be retried. :param timeout: The amount of time, in seconds, to wait for the request to complete. Note that if ``retry`` is specified, the timeout applies to each individual attempt. :param metadata: Additional metadata that is provided to the method. :param gcp_conn_id: The connection ID to use connecting to Google Cloud. :param impersonation_chain: Optional service account to impersonate using short-term credentials, or chained list of accounts required to get the access_token of the last account in the list, which will be impersonated in the request. If set as a string, the account must grant the originating account the Service Account Token Creator IAM role. If set as a sequence, the identities from the list must grant Service Account Token Creator IAM role to the directly preceding identity, with first account from the list granting this role to the originating account (templated). """
[docs] template_fields: Sequence[str] = ("operation_name", "region", "project_id", "impersonation_chain")
def __init__( self, *, operation_name: str, region: str, project_id: str = PROVIDE_PROJECT_ID, retry: Retry | _MethodDefault = DEFAULT, timeout: float | None = None, metadata: Sequence[tuple[str, str]] = (), gcp_conn_id: str = "google_cloud_default", impersonation_chain: str | Sequence[str] | None = None, **kwargs, ): super().__init__(**kwargs) self.operation_name = operation_name self.region = region self.project_id = project_id self.retry = retry self.timeout = timeout self.metadata = metadata self.gcp_conn_id = gcp_conn_id self.impersonation_chain = impersonation_chain
[docs] def execute(self, context: Context): hook = DataprocHook(gcp_conn_id=self.gcp_conn_id, impersonation_chain=self.impersonation_chain) self.log.info("Canceling operation: %s", self.operation_name) hook.get_operations_client(region=self.region).cancel_operation(name=self.operation_name) self.log.info("Operation canceled.")

Was this entry helpful?