Closed smira closed 2 years ago
So this changes to mark control plane as initialized as soon as the workload Kubernetes endpoint responds to the request to allow CAPI core controllers to set up proper workload cluster watches
the actual fix should be https://github.com/kubernetes-sigs/cluster-api/pull/6425, but this PR is a small step in the right direction
/m
This should resolve a problem when there's a "deadlock" between CAPI core MachineController and CACPPT: CAPI core waits for the control plane to be initialized to start watching workload cluster Nodes to pick up
.spec.ProviderID
changes, while CACPPT relies on the MachineController to update the Machine status.Signed-off-by: Andrey Smirnov andrey.smirnov@talos-systems.com