dimatha
commented
2 years ago Ok, I had to spend some time to understand it. And here is my experience.
From the proposal:
Region -> ComputeCluster, Zone -> HostGroup
#no tag required
#hostgroups need to be pre-configured
#CAPV need permissions to create vm groups and affinity rulesFrom my experience:
- Create vCenter Tags and categories
- Create Host Groups
- Create VM groups with a dummy VM as it can't be empty
- Create VM to Host rules
- Create CAPV resources: VSphereDeploymentZone => VSphereFailureDomain
- make sure VSphereDeploymentZone is ready
- no need to reference VSphereDeploymentZone from the CP
- assigned zone should be visible from the Machine spec.failureDomain
- all ready VSphereDeploymentZones should be visible from the Cluster status.failureDomains
gab-satchi
srm09
k8s-triage-robot
/kind feature
Describe the solution you'd like As a cluster admin I'd like to distribute k8s CP nodes across different hosts within the same cluster.
Referring to this proposal: https://github.com/kubernetes-sigs/cluster-api-provider-vsphere/blob/master/docs/proposal/20201103-failure-domain.md#for-user-story-1 I'm struggling to understand how to do the configuration using VSphereFailureDomain CRD to achieve the goal.
Anything else you would like to add: [Miscellaneous information that will assist in solving the issue.]
Environment:
kubectl version):/etc/os-release):