Search code examples
google-cloud-platformairflowsftpgoogle-cloud-networking

How to perform SFTP to a remote server using airflow

We have a scenario where we need to copy a file from GCS (google cloud storage) to a remote SFTP server via airflow without using any intermediate on-premise Unix server.

Is there any way to achieve this task with/without using GCP Compute Service or using docker POD via airflow ?

Solution

  • Yes, you can use GCSToSFTPOperator:

    from airflow.providers.google.cloud.transfers.gcs_to_sftp import GCSToSFTPOperator

copy_file_from_gcs_to_sftp = GCSToSFTPOperator(
    task_id="file-copy-gsc-to-sftp",
    sftp_conn_id=SFTP_CONN_ID,
    source_bucket=os.environ.get("GCP_GCS_BUCKET_1_SRC", "test-gcs-sftp"),
    source_object="file.csv",
    destination_path="/tmp/single-file/",
)

    The operator downloads the file to temp dir on the Airflow worker and upload it to the SFTP server. You can read more about the operator in the docs.