feat: support keystone-auth

[okozachenko1203]

fix #9

[okozachenko1203]

This has been tested on local lab

[mnaser]

I think we can take this as an opportunity to start cleaning up all of our manifests which are bundled. I think it's very difficult to manage them at the moment and I would like to suggest the following:

1. Create a charts folder and start bundling the charts in there, if there is no chart, we build it (in this case for this)
2. Since we already have Helm on the system, we can simply use helm template with a custom values file instead of using Jinja2
3. With that in place, we're able to simply just bump the chart data in here and easily change the overrides by relying on values instead of raw manifest files.
4. The templated value is what we "feed" into the ClusterResourceSet, and it also makes it easier for the long run to transition to the addon providers as well since we will have the charts ahead of time.

Can you make changes to this to accomodate for that?

[okozachenko1203]

I think we can take this as an opportunity to start cleaning up all of our manifests which are bundled. I think it's very difficult to manage them at the moment and I would like to suggest the following:

1. Create a charts folder and start bundling the charts in there, if there is no chart, we build it (in this case for this)
2. Since we already have Helm on the system, we can simply use helm template with a custom values file instead of using Jinja2
3. With that in place, we're able to simply just bump the chart data in here and easily change the overrides by relying on values instead of raw manifest files.
4. The templated value is what we "feed" into the ClusterResourceSet, and it also makes it easier for the long run to transition to the addon providers as well since we will have the charts ahead of time.

Can you make changes to this to accomodate for that?

what if i create a helm chart and deploy like cluster-autoscaler instead of rendering templates and feeding into ClusterResourceSet configmap?

[mnaser]

@okozachenko1203: this assumes that magnum-conductor can reach the cluster directly. In Atmosphere, it can. However, when users are using OSA or Kolla, that is likely not the case, so we'll have to rely on CAPO for the cluster to be reachable (especially in scenarios when the cluster proxy is running).

Done