kubernetes-sigs / cluster-api

Home for Cluster API, a subproject of sig-cluster-lifecycle

Home Page:https://cluster-api.sigs.k8s.io

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ClusterClass MachinePool implementation is probably not able to rollout BootstrapConfig changes.

sbueringer opened this issue · comments

Quoting from a different issue:

Can you please summarize what a user would have to do to rollout a change to a KubeadmConfig with a reguar cluster (not using ClusterClass). Is it as simple as modifying the KubeadmConfig object that is linked in a MachinePool directly?

Change MachinePool.spec.template.spec.bootstrap.configRef.name and the KubeadmConfig.spec changes will be reconciled.
In practice, that means a user's KubeadmConfig.metadata.name should be suffixed with a hash (or increasing number) when making changes, and the reference in the MachinePool object gets changed to this new KubeadmConfig name.
#8858 (comment)

The ClusterClass MachinePool implementation today uses one single BootstrapConfig / KubeadmConfig object for a MachinePool and just continuously patches it in-place to rollout changes.

Based on #8858 this is not enough, it seems to be required to rotate the BootstrapConfig object (create a new one + update the ref in MachinePool).

Someone would need to confirm if my observation is correct.

Furthermore I don't know if there is a documented contract as of today for MachinePools how BootstrapConfigs are supposed to be rolled out. I also don't know if MachinePools behave the same across providers today.

So my tl;dr:

  • ClusterClass can probably not rollout BootstrapConfig changes for MachinePools today
  • We need a documented contract on how BootstrapConfig rollouts are supposed to work with MachinePools

This issue is currently awaiting triage.

CAPI contributors will take a look as soon as possible, apply one of the triage/* labels and provide further guidance.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

I don't have the know how about MachinePools and MachinePool implementations in providers to be 100% sure. I also don't have the time to do the required research, but I just wanted to surface this issue, so folks have a chance to fix it (or determine that everything is actually fine at the moment).

(cc @AndiDog @willie-yao, just fyi)

/priority important-longterm
/kind bug