Closed mxmxchere closed 5 months ago
We can take the current helm values.yaml as a starting point for the cluster-class variables. Currently we have the following values(now)-variables(then) in the helmchart:
variable | Comment | Status |
---|---|---|
external_id | probably not needed | ✅ |
controller_flavor | done | ✅ |
worker_flavor | done | ✅ |
node_cidr | done | ✅ |
dns_nameservers | done | ✅ |
openstack_security_groups | done | ✅ |
prefix | helm templating only | 🟠 |
testcluster_name | helm only | 🟠 |
control_plane_machine_gen | helm only | 🟠 |
worker_machine_gen | helm only | 🟠 |
etcd_unsafe_fs | always false, no need to template | ✅ |
http_proxy | tbd | ❌ |
no_proxy | tbd | ❌ |
containerd_registry_files | tbd | ❌ |
etcd_defrag | always true, no need to template | ✅ |
tweak_kubeapi_memlimit | always true, no need to template | ✅ |
A 1:1 translation of "helm-templating" to "cluster-class-patching" will not be possible everywhere.
prefix
and testcluster_name
are a little bit tricky for the following reasons:
all helm values are included as cluster class variables now, creating the patches is the next step
As mentioned in https://github.com/SovereignCloudStack/k8s-cluster-api-provider/issues/666 the helm-chart that reflects the current state of https://github.com/SovereignCloudStack/k8s-cluster-api-provider now exists. Because the cluster stack operator can not be used to template a helm chart we will need to reimplement the templating using the clusterclass templating mechanism.