fix: Increase robustness of openshift login flow

vburckhardt commented 3 years ago

This PR increases the robustness of the openshift login flow:

Retries on retrieving the passcode ( including covering of the rate limit that kicks in)
Race condition (around request headers) that existed in previous version as the same rest client instance was reused without proper locking. Some threads in previous code was effectively essentially removing the auth headers causing any parallel and subsequent request to fail with 401.
Exiting with error when passcode cannot be retrieved - ensuring the user is not unexpectedly authenticated as 'anonymous'
Handling of intermittent issue where openshift returns a 200 instead of a 302 with accesscode on authentication flow

Context: We've been plagued with a number of authentication / login issue in our terraform code. The IBM Cloud terraform provider makes intensive use. See in particular https://github.com/IBM-Cloud/terraform-provider-ibm/issues/2811

hkantare commented 3 years ago

We will looking into the PR

ocofaigh commented 3 years ago

@hkantare Vincent is on vacation for 2 weeks, so I am just following up on this. These authentication issues are occurring quite frequently, so would be great if we could add some robustness like Vincent has proposed in this PR. Thank you.

hkantare commented 3 years ago

Thanks we reviewd the PR and all looks good

ocofaigh commented 3 years ago

@hkantare Thank you very much - How do we go about getting this into the ibm terraform provider code for next provider release? Thanks

IBM-Cloud / bluemix-go

fix: Increase robustness of openshift login flow #318