feat: support auto-retry of failed network requests to `/oauth/token` endpoint

evansims commented 8 months ago

Description

This pull request:

Introduces support for auto-retries of failed HTTP network requests (429 and 5xx-series responses) when communicating with the /oauth/token authentication endpoint.
Adds a jitter function to supply exponential backoff values for retries.
Updates and adds new tests to cover the changes.

This PR also bumps the pytest-cov dependency from >= 4.1.0, < 5 to >= 5, < 6, as no changes were required for our tests to support the latest version.

References

Closes https://github.com/openfga/python-sdk/issues/18

Review Checklist

[x] I have clicked on "allow edits by maintainers".
[ ] I have added documentation for new/changed functionality in this PR or in a PR to openfga.dev [Provide a link to any relevant PRs in the references section above]
[x] The correct base branch is being used, if not main
[x] I have added tests to validate that the change in functionality is working as expected

evansims commented 8 months ago

@rhamzeh I'm currently re-using the existing max_retry and min_wait_in_ms configuration values (that we're using for FGA retries) to configure this auth retry strategy.

Do you think this is appropriate? Should we be adding new configuration properties?

rhamzeh commented 8 months ago

Do you think this is appropriate? Should we be adding new configuration properties?

No, reusing is fine. It's what we do in the other SDKs.

evansims commented 8 months ago

Updated based on feedback from https://github.com/openfga/sdk-generator/pull/335

openfga / python-sdk