This message was deleted.

I use this

kubeconfig

in a

k8s.Provider

created in the code. This provider is used for a number of downstream resources. The change in

kubeconfig

output leads to the provider being `replace`d, which leads to all these resources being replaced…

@gorgeous-egg-16927?

Without knowing the details, I’d guess that this is just an overly conservative preview. If the

kubeconfig

changes and includes a computed value, the preview can sometimes not determine if the cluster has changed, which would require the resources to be replaced. A workaround is to update the cluster separately from the resources deployed to it so that the

kubeconfig

is known during preview.