Add RefreshError to retryable errors

[baeminbo]

Thanks for stopping by to let us know something could be better!

PLEASE READ: If you have a support contract with Google, please create an issue in the support console instead of filing on GitHub. This will ensure a timely response.

Is your feature request related to a problem? Please describe.
RefreshError can be raised if GCE metadata server returns 500 error for service account token. But, we currently retry only for TransportError at retry.py.

This makes GCP libraries depending on the default retry fail by the transient auth error.

Describe the solution you'd like
Add auth_exceptions.RefreshError to if_transient_error.

Describe alternatives you've considered
Core API users can use a custom retry that handles RefreshError as retryable.

Additional context
See this screenshot of a RefreshError example that happened at GCS API in a Dataflow job.

[tswast]

Duplicate of #223 ?

[baeminbo]

IIUC, #223 is about Unauthenticated by 401 from metadata server, which is caused by time sync. But, this is an issue with 500 error from metadata server. So, I think they are different issues.

[liu-du]

Hi,

I'm experiencing the same transient issues recently but in my case metadata server returns 404 not found. The error is:
google.auth.exceptions.RefreshError: ("Failed to retrieve http://metadata.google.internal/computeMetadata/v1/instance/service-accounts/default/identity?audience=REDACTED&format=full from the Google Compute Engine metadata service. Status: 404 Response:\nb'Not Found\n'", <google.auth.transport.requests._Response object at 0x7f44b3dcf8d0>)

[parthea]

Retry logic was added in googleapis/google-auth-library-python#1113. In the case of the 500 errors, there is an open question in googleapis/google-auth-library-python#984 requesting more information about how often they are occurring. Please open a new issue in https://github.com/googleapis/google-auth-library-python as that is where the retry logic should be improved.