Simplify controller architecture, merge "operator" and "controller"

[ericavonb]

The operator controller for deploying the core controller is an extraneous layer. The "controller" is the operator, it handles the operational logic specific to this operator (the machine configuration). Other operators exist to handle deploying the operators themselves (CVO in this case,.

[cgwalters]

I agree with this - although with this type of thing I have a temptation to ping Abhinav to understand more of his reasoning, but OTOH he's pretty busy with things like reviewing a gigantic PR to add baremetal to the installer, etc.

Probably makes the most sense to approach this in incremental phases; the fact that we have just one image now is a big one! Next would be to run the operator and controller in the same pod probably.

After that we can start to take advantage of that and drop e.g. the controllerconfig CRD which I think is only used as a way to sync from the operator ➡️ controller...we could just use a yaml file in the pod shared memory or whatever.

And then perhaps after that we could link the controller's code into the operator process.

[abhinavdahiya]

I disagree with operator is the controller.

The controller + server + daemon is the ecosystem that realizes the machineconfiguration API, provides a way to manage configuration of machines in the cluster.

The operator is the one that knows information about how to configure them and update them through versions.

For example,
configpool is transparent to controller, it updates the nodes with new configuration with max unavail set. The operator knows a configpool is for control plane so it should be set to something safe.

Or that when upgrading it's required that control plane upgrades while the compute ones can trail..

I think the management and control-loop layer separation is useful.

[openshift-bot]

Issues go stale after 90d of inactivity.

Mark the issue as fresh by commenting /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
Exclude this issue from closing by commenting /lifecycle frozen.

If this issue is safe to close now please do so with /close.

/lifecycle stale

[openshift-bot]

Stale issues rot after 30d of inactivity.

Mark the issue as fresh by commenting /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.
Exclude this issue from closing by commenting /lifecycle frozen.

If this issue is safe to close now please do so with /close.

/lifecycle rotten
/remove-lifecycle stale

[abhinavdahiya]

/lifecycle frozen