Matthew Booth
Matthew Booth
> We need [kubernetes-sigs/cluster-api-provider-openstack#2096](https://github.com/kubernetes-sigs/cluster-api-provider-openstack/pull/2096) first. Can you put this dependency in the description as a TODO item?
@jfcavalcante are you still interested in this? Seems promising.
How do you see client code using this? Here's an example of how we're using mocks in CAPO: https://github.com/kubernetes-sigs/cluster-api-provider-openstack/blob/852272197ad100fecd9cc7e95a77e88897812520/pkg/cloud/services/compute/instance_test.go#L82-L96 ```golang { testName: "Return image ID when name given", image: infrav1.ImageParam{...
I've been thinking about this since I first raised it a few months ago. As implemented, the apiVersion field is a status field, not a spec field, because it can't...
> lgtm for the API from my side (+/- the godoc nit) > > Before I start to review the implementation, @fabriziopandini @chrischdi @mdbooth @lentzi90 Are we fine with the...
Coming back to this in the morning, I think I can summarise the issue more succinctly: MachineSpec, and therefore MachineTemplateSpec, refers to a failure domain only by name. It is...
CAPV appears to suffer with the same problem. AFAICT updating a `VSphereDeploymentZoneSpec`, for example with a new `PlacementConstraint`, would not result in a rollout of new machines with the new...
CAPV's `Topology` looks very similar to what we're proposing in OpenStack: https://github.com/kubernetes-sigs/cluster-api-provider-vsphere/blob/9cb1cec4cf7206358ae9934082a922b955a504f1/apis/v1beta1/vspherefailuredomain_types.go#L60-L82 I have still not found if/how an update to this object would be reflected in existing machines.
/area provider/control-plane-kubeadm /area machineset /area machinedeployment
> Can we include some use cases that are impacted/improved by this? I have 'corporate naming policy' which accounts for a great many sins in practise. Such policy is often...