Copilot commented on code in PR #69190: URL: https://github.com/apache/airflow/pull/69190#discussion_r3560509614
########## providers/amazon/tests/system/amazon/aws/example_s3_compatible_object_storage.py: ########## @@ -0,0 +1,98 @@ +# +# 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. +""" +Example Dag: input -> transform -> output on an S3-compatible object store via ``ObjectStoragePath``. + +The Amazon provider talks to any S3-compatible object store, so ``ObjectStoragePath("s3://...")`` +reaches the store through the amazon provider once the ``aws`` connection points at its S3 +endpoint. Amazon S3 is the baseline; the same code works against other S3-compatible services Review Comment: In the module docstring, “amazon provider” is a proper noun and should be capitalized consistently as “Amazon provider”. ########## providers/amazon/tests/system/amazon/aws/example_s3_compatible_object_storage.py: ########## @@ -0,0 +1,98 @@ +# +# 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. +""" +Example Dag: input -> transform -> output on an S3-compatible object store via ``ObjectStoragePath``. + +The Amazon provider talks to any S3-compatible object store, so ``ObjectStoragePath("s3://...")`` +reaches the store through the amazon provider once the ``aws`` connection points at its S3 +endpoint. Amazon S3 is the baseline; the same code works against other S3-compatible services +(for example Amazon S3, Backblaze B2, Cloudflare R2, and MinIO). See the recipe "Use an +S3-compatible object store for Airflow remote task logs" for connection and ``[logging]`` setup. + +Set up an ``aws`` connection (default id ``aws_s3``) whose ``extra`` includes the S3 +``endpoint_url`` and ``region_name``. The bucket name comes from ``S3_BUCKET_NAME``. + +Requires the s3fs extra: ``pip install 'apache-airflow-providers-amazon[s3fs]'``. +""" + +from __future__ import annotations + +import os +from datetime import datetime + +from airflow.sdk import ObjectStoragePath, dag, task + +DAG_ID = "example_s3_compatible_object_storage" + +# Connection id and bucket are read from the environment so the example carries no secrets. +S3_CONN_ID = os.environ.get("S3_CONN_ID", "aws_s3") +S3_BUCKET_NAME_PLACEHOLDER = "replace-with-your-s3-bucket" +S3_BUCKET_NAME = os.environ.get("S3_BUCKET_NAME", S3_BUCKET_NAME_PLACEHOLDER) +BASE_URI = f"s3://{S3_CONN_ID}@{S3_BUCKET_NAME}/airflow-demo/" + + +def _base_path() -> ObjectStoragePath: + if S3_BUCKET_NAME == S3_BUCKET_NAME_PLACEHOLDER: + raise ValueError("Set S3_BUCKET_NAME to a real bucket name before running this Dag.") + return ObjectStoragePath(BASE_URI) + + +@dag( + schedule=None, + start_date=datetime(2021, 1, 1), + catchup=False, + tags=["example", "s3-compatible", "object-storage"], +) +def example_s3_compatible_object_storage(): + """Write input to the object store, transform it, and write the output back.""" + + @task + def input_to_store() -> str: + """Write a raw input object to the store and return its path.""" + base = _base_path() + base.mkdir(exist_ok=True) + src = base / "input.txt" + src.write_text("s3\ncompatible\nobject\nstorage\n") + return str(src) + + @task + def transform(src_path: str) -> str: + """Read the input from the store, uppercase it, and write the result back.""" + src = ObjectStoragePath(src_path) + text = src.read_text() + base = _base_path() + dst = base / "output.txt" + dst.write_text(text.upper()) + return str(dst) + + @task + def output_from_store(dst_path: str) -> None: + """Read the transformed object back from the store to confirm the round-trip.""" + dst = ObjectStoragePath(dst_path) + print(dst.read_text()) + + output_from_store(transform(input_to_store())) + + +dag = example_s3_compatible_object_storage() + + +from tests_common.test_utils.system_tests import get_test_run # noqa: E402 + +# Needed to run the example DAG with pytest (see: contributing-docs/testing/system_tests.rst) +test_run = get_test_run(dag) Review Comment: `get_test_run(dag)` executes this Dag via `dag.test()` and asserts success. With the current default `S3_BUCKET_NAME` placeholder, `_base_path()` raises `ValueError`, so this system-test module will fail whenever it is executed without explicitly setting `S3_BUCKET_NAME`. To keep the example Dag importable while preventing accidental failures in the system-test suite, mark `test_run` as skipped when `S3_BUCKET_NAME` is left at the placeholder value (or otherwise unset). ########## providers/amazon/docs/logging/s3-compatible-remote-logging.rst: ########## @@ -0,0 +1,211 @@ + .. 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. + +.. _write-logs-s3-compatible: + +Use an S3-compatible object store for Airflow remote task logs +============================================================== + +The Amazon provider talks to any S3-compatible object store, not just Amazon S3. Because the +:ref:`S3 remote task handler <write-logs-amazon-s3>` issues standard S3 API calls, pointing the +``aws`` connection at a custom ``endpoint_url`` makes it write Airflow task logs to that +endpoint with no new provider and no core change. You use the ``s3://`` scheme in +``[logging]`` exactly as you would for Amazon S3, and the same connection also backs +``ObjectStoragePath("s3://...")`` for Dag data. + +Amazon S3 is the baseline. The same steps work against other services that expose an +S3-compatible API, for example Amazon S3, Backblaze B2, Cloudflare R2, and MinIO. The only +per-provider differences are the endpoint URL, the region, and whether path-style addressing +is required. + +This recipe targets Airflow 3.x with ``apache-airflow-providers-amazon``. + +Prerequisites +------------- + +- A bucket for logs (private). The examples use ``$S3_BUCKET_NAME``. +- An access key and secret scoped to that bucket. Prefer a bucket-scoped key over an + account-wide one. +- ``apache-airflow-providers-amazon`` installed. For ``ObjectStoragePath`` you also need the + ``s3fs`` extra: ``pip install 'apache-airflow-providers-amazon[s3fs]'``. + +Every S3-compatible service issues an access key id and a secret access key. Map them onto the +AWS connection fields as follows. + +============================ ================================= ============================================= +S3-compatible value Standardized env var AWS connection field +============================ ================================= ============================================= +Access key id ``S3_ACCESS_KEY_ID`` ``login`` (AWS access key id) +Secret access key ``S3_SECRET_ACCESS_KEY`` ``password`` (AWS secret access key) +Bucket name ``S3_BUCKET_NAME`` used in ``remote_base_log_folder`` +Region ``S3_REGION`` ``extra.region_name`` +S3 endpoint ``S3_ENDPOINT`` ``extra.endpoint_url`` +============================ ================================= ============================================= + +Find the endpoint and region for your bucket in your provider's console or CLI. For Amazon S3 +the endpoint is the default AWS endpoint and you can omit ``endpoint_url`` entirely; for other +S3-compatible services set ``endpoint_url`` to the provider's S3 endpoint, such as +``https://your-s3-endpoint.example.com``. The connection ``extra.endpoint_url`` must include a +scheme, for example ``https://`` or ``http://``. + +Step 1: Create the connection pointing at your endpoint +------------------------------------------------------- + +Create an ``aws`` connection whose ``endpoint_url`` extra is your S3 endpoint. The amazon +provider sends every S3 call to that endpoint instead of the AWS default, which is what makes +the S3 handler talk to your store. For Amazon S3 you can leave ``endpoint_url`` unset and the Review Comment: Capitalization: “Amazon provider” is a proper noun, so this should be “Amazon provider” (matching earlier usage in the same doc). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
