[Question] Should yurt-controller-manager deploy as Daemonset or static pod on control plane?

[DrmagicE]

What happened:
yurtctl/yurtadm use 1 replicas Deployment to deploy yurt-controller-manager. Under such circumstances, if the node where yurt-controller-manager deployed shutdowns, the node won't turn into NotReady state, so the yurt-controller-manager won't be re scheduled which may cause issues.

What you expected to happen:
As the yurt-controller-manager is a replacement of the NodeLifecycleController in kube-controller-manager, which is usually
deployed as a static pod on each control plane node.

Maybe the yurt-controller-manager should follow the same deployment way (Daemonset or static pod)?

How to reproduce it (as minimally and precisely as possible):

Anything else we need to know?:

Environment:

• OpenYurt version:
• Kubernetes version (use kubectl version):
• OS (e.g: cat /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

others
/kind question

[rambohe-ch]

@DrmagicE Thanks for raising the issue. At present we can use deployment.replicas=2 in order to make the ycm high availbility.

btw: we are now reforming ycm, remove NodeLifeCycle controller from ycm and put NodeLifeCycle controller back to kcm.
issue reference link: #776

[DrmagicE]

@rambohe-ch Thanks for your quick response. I would try daemonset for now and looking forward to the new feature.