From 7b643fb27208dfe0195ecee68e4b33d62cde951c Mon Sep 17 00:00:00 2001 From: nlarge-google Date: Thu, 7 Nov 2024 16:58:11 +0000 Subject: [PATCH] Fix: Migrate NOAA Passive Bioacoustic DAG to new environment --- .../noaa_passive_bioacoustic_dag.py | 10 +++------- .../pipelines/noaa_passive_bioacoustic/pipeline.yaml | 10 +++------- 2 files changed, 6 insertions(+), 14 deletions(-) diff --git a/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/noaa_passive_bioacoustic_dag.py b/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/noaa_passive_bioacoustic_dag.py index e1f951105..15364f233 100644 --- a/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/noaa_passive_bioacoustic_dag.py +++ b/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/noaa_passive_bioacoustic_dag.py @@ -49,8 +49,9 @@ task_id="kub_gcs_to_bq", startup_timeout_seconds=1000, name="load_data", - namespace="composer", - service_account_name="datasets", + namespace="composer-user-workloads", + service_account_name="default", + config_file="/home/airflow/composer_kube_config", image_pull_policy="Always", image="{{ var.json.noaa_passive_bioacoustic.container_registry.run_csv_transform_kub }}", env_vars={ @@ -62,11 +63,6 @@ "SCHEMA_FILEPATH": "schema.json", "HEADER": '["CLOUD_PATH" ,"FILE_NAME" ,"START_DATE" ,"START_TIME" ,"DATA_COLLECTION_NAME" ,"SOURCE_ORGANIZATION" ,"FUNDING_ORGANIZATION" ,"PROJECT_NAME" ,"SENSOR_DEPTH" ,"SAMPLE_RATE" ,"DURATION" ,"PLATFORM_NAME" ,"SHAPE1" ,"SHAPE2" ,"SHAPE3" ,"SHAPE4" ,"SHAPE5" ,"SHAPE6" ,"SHAPE7" ,"SHAPE8" ,"SHAPE9" ,"SHAPE10" ]', }, - resources={ - "request_memory": "2G", - "request_cpu": "1", - "request_ephemeral_storage": "10G", - }, ) gcs_to_gcs_operator >> kub_gcs_to_bq diff --git a/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/pipeline.yaml b/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/pipeline.yaml index d1eaa5193..5502fe3f6 100644 --- a/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/pipeline.yaml +++ b/datasets/noaa_passive_bioacoustic/pipelines/noaa_passive_bioacoustic/pipeline.yaml @@ -48,8 +48,9 @@ dag: task_id: "kub_gcs_to_bq" startup_timeout_seconds: 1000 name: "load_data" - namespace: "composer" - service_account_name: "datasets" + namespace: "composer-user-workloads" + service_account_name: "default" + config_file: "/home/airflow/composer_kube_config" image_pull_policy: "Always" image: "{{ var.json.noaa_passive_bioacoustic.container_registry.run_csv_transform_kub }}" env_vars: @@ -62,10 +63,5 @@ dag: HEADER: >- ["CLOUD_PATH" ,"FILE_NAME" ,"START_DATE" ,"START_TIME" ,"DATA_COLLECTION_NAME" ,"SOURCE_ORGANIZATION" ,"FUNDING_ORGANIZATION" ,"PROJECT_NAME" ,"SENSOR_DEPTH" ,"SAMPLE_RATE" ,"DURATION" ,"PLATFORM_NAME" ,"SHAPE1" ,"SHAPE2" ,"SHAPE3" ,"SHAPE4" ,"SHAPE5" ,"SHAPE6" ,"SHAPE7" ,"SHAPE8" ,"SHAPE9" ,"SHAPE10" ] - resources: - request_memory: "2G" - request_cpu: "1" - request_ephemeral_storage: "10G" - graph_paths: - "gcs_to_gcs_operator >> kub_gcs_to_bq"