x-snap-config: Do not check if node is part of the cluster

[bschimke95]

Follow-up fix for: #633

In the mentioned PR, we wait until k8sd is up to sync the config. We exit early if the node is not part of a cluster. However, a closer look at the implementation of NodeStatus shows that this would skip any error that we wait for:

func (c *k8sd) NodeStatus(ctx context.Context) (apiv1.NodeStatusResponse, bool, error) {
	response, err := query(ctx, c, "GET", apiv1.NodeStatusRPC, nil, &apiv1.NodeStatusResponse{})
	if err != nil {
		// Error 503 means the node is not initialized yet
		var statusErr api.StatusError
		if errors.As(err, &statusErr) {
			if statusErr.Status() == http.StatusServiceUnavailable {
				return apiv1.NodeStatusResponse{}, false, nil
			}
		}

		return apiv1.NodeStatusResponse{}, false, err <- Returns partOfCluster == false if an error happens. 
	}

	return response, true, nil
}

This PR removes the wrong check.

[addyess]

So, if NodeStatus() always returns partofCluster=false for any error why do we no longer need this exit early?

i'm not sure i have the context from #633 to properly review this. Maybe @mateoflorido

[addyess]

basically you're saying any error from NodeStatus at all exits early-- and that's wrong?

[bschimke95]

So, WaitUntilReady will retry if the bool return value is false. If something in the error return value is returned, then this exits immediately.

NodeStatus second return value indicates if the node is part of a cluster or not. However, it will also return false for this argument if an error is returned (third return value). The current implementation would always exit with Node is notpart of cluster even if NodeStatus returned and error (which means the "is part of cluster" return value is not valid).

[addyess]

AHH! i didn't look down at lines 70-71. Thank you

[addyess]

LGTM, thanks for the explanation!