ugh yikes, just ran into total failure trying to perform an update that required replacing an EKS cluster

error: post-step event returned an error: failed to save snapshot: .pulumi/stacks/base.non-prod.json: snapshot integrity failure; it was already written, but is invalid (backup available at .pulumi/stacks/base.non-prod.json.bak): resource urn:pulumi:base.non-prod::yd-base::yd:eks:Cluster$eks:index:Cluster$kubernetes:core/v1:ConfigMap::yd-eks-nodeAccess refers to unknown provider urn:pulumi:base.non-prod::yd-base::yd:eks:Cluster$eks:index:Cluster$pulumi:providers:kubernetes::yd-eks-eks-k8s::b48893e2-fd53-4bed-9753-c28de972084c

i think i understand the error (the provider was deleted when the cluster was replaced), but unsure how to resolve restoring the backup, any tips?

we're sorry for this experience. generally, if you use autonaming and don't provide explicit clusternames, this doesn't happen the reason it happens when using explicit names is because the provider resource has an explicit name in it, the cluster disappears and then there's nowhere to put the the kubernetes resources

yeah, okay, that makes sense. i understand the issue, and generally this kind of operation isn’t the sort of thing i’d expect to do in production, just didn’t expect to blow my foot off either

Thanks for leaving a comment on https://github.com/pulumi/pulumi/issues/4996, Jem, and sorry you hit this. I think that with the experience you’ve described in the thread we may have enough data to repro this, which has been tricky in the past.

a little more info, the biggest clean up was helm chart objects, which even though they expressed a parent relationship to the cluster, got left behind