jceresini opened a new issue #21440:
URL: https://github.com/apache/airflow/issues/21440
### Apache Airflow Provider(s)
google
### Versions of Apache Airflow Providers
$ pip freeze | grep airflow-providers
apache-airflow-providers-apache-beam==3.1.0
apache-airflow-providers-ftp==2.0.1
apache-airflow-providers-google==6.3.0
apache-airflow-providers-http==2.0.2
apache-airflow-providers-imap==2.1.0
apache-airflow-providers-sqlite==2.0.1
### Apache Airflow version
2.2.3 (latest released)
### Operating System
MacOS 11.6.1 (20G224)
### Deployment
Virtualenv installation
### Deployment details
The only python packages explicitly installed were
`apache-airflow-providers-google` and `apache-airflow-providers-apache-beam`.
### What happened
[2022-02-08 16:44:23,610] {credentials_provider.py:330} INFO - Getting
connection using `google.auth.default()` since no key file is defined for hook.
[2022-02-08 16:44:25,409] {http.py:123} WARNING - Invalid JSON content from
response: b'{\n "error": {\n "code": 403,\n "message": "User must be
authenticated when user project is provided",\n "status":
"PERMISSION_DENIED",\n "details": [\n {\n "@type":
"type.googleapis.com/google.rpc.ErrorInfo",\n "reason":
"USER_PROJECT_DENIED",\n "domain": "googleapis.com",\n
"metadata": {\n "service": "dataflow.googleapis.com",\n
"consumer": "projects/cd-np-test1"\n }\n }\n ]\n }\n}\n'
That first line is expected. The operator should be calling
`google.auth.default()` to get my credentials. With some quick debugging I
confirmed that is in fact the case, and it has the correct credentials.
After that, retrieving the discovery document fails with a 403. If I force
the api client to use a cached copy of the discovery document, the task
completes just fine. So the authentication is working when making calls to
create the dataflow job. But for some reason the dataflow API's discovery
document url doesn't like the request. Note that you can access the discovery
document with no auth at all:
https://www.googleapis.com/discovery/v1/apis/dataflow/v1b3/rest
### What you expected to happen
I should be able to use local application default credentials with running
airflow locally, to trigger the dataflow operator
### How to reproduce
Ran the following in a new venv:
```
gcloud auth application-default login # successfully authenticated as my user
python3 -m pip install apache-airflow-providers-google
apache-airflow-providers-apache-beam
export AIRFLOW__CORE__LOAD_EXAMPLES=false
export AIRFLOW_HOME=$(pwd)/airflowhome
airflow db init
# Copy the python file into the ./airflowhome/dags (below)
airflow tasks test test_dag dataflow_task $(date +%Y-%m-%dT%H:%M:%S)
```
Contents of the python file:
```python
from airflow.providers.google.cloud.operators.dataflow import (
DataflowStartFlexTemplateOperator,
)
from datetime import datetime
from airflow import DAG
with DAG(dag_id="test_dag", schedule_interval=None,
start_date=datetime.now()) as dag:
dataflow_task = DataflowStartFlexTemplateOperator(
task_id="dataflow_task",
body={
"launchParameter": {
"containerSpecGcsPath": "gs://foo/bar/baz.json",
"jobName": "test",
"environment": {
"serviceAccountEmail":
"[email protected]",
},
}
},
wait_until_finished=True,
location="us-central1",
project_id="some-project",
)
```
### Anything else
The `google-api-python-client` is locked to versions `<2.0`. In newer
versions of the client, discovery documents are shipped with the client and the
lookup is rarely used. As long as upgrading that dependency doesn't break other
things, that might be a simple fix.
### Are you willing to submit PR?
- [ ] Yes I am willing to submit a PR!
### Code of Conduct
- [X] I agree to follow this project's [Code of
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
--
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]