Wrong openstack provider about afterburn HOT 4 CLOSED

@lgogolin thanks for the report.
While I understand this may be surprising, it is working as expected. In particular, you are hitting #251 (which still need to be part of some proper documentation somewhere).

For your specific case, OpenStack in general does not mandate a metadata-service as the baseline. Thus you can be in either one of these two cases:

1. your deployment does NOT have a metadata service -> do not run afterburn, it can't fetch/write attributes anyway
2. your deployment has a metadata service -> write a dropin for afterburn service with Environment=AFTERBURN_OPT_PROVIDER=--provider openstack-metadata

from afterburn.

@lucab Yeah, I make it work by injecting parameters and forcing openstack-metdata as provider.

It was more about IMO wrong name for openstack provider hardcoded in the app which makes default approach --cmdline not working out of the box.

If You think this is not a BUG - We can close the issue

from afterburn.

was more about IMO wrong name for openstack provider hardcoded in the app which makes default approach --cmdline not working out of the box.

I agree this is a terrible UX, and the bug is that it should be properly documented.

The root of this is that FCOS images can work with or without a metadata endpoint (thus the generic openstack platform ID), but Afterburn cannot (thus the -metadata suffix).

Sorry again for the pain/friction, I'm going to close this ticket and hopefully sometime soon we will have an end-to-end "Booting FCOS on OpenStack" docpage.

from afterburn.

I read the issue You mentioned #251 and also issue from openshift as reference.

And to be honest do not understand where is the problem.

You mentioned there that metadata service in OpneStack is optional - true
I just do not see difference in approaches:

--cmdline does not work because ignition.id does not match
Fixed ignition.id but metadata not available - it will also fail on requesting data.

from afterburn.