Option to suppress output of data sources defined within check blocks during plans

Terraform
1

We’re interesting in rolling out check blocks to verify various life states of our Kubernetes components. We’re considering using check blocks for this, with the data source kubernetes_resource to pull in life state.

This works well, except that the extremely large json object is printed during the plan for each of these data sources:

 # module.karpenter.data.kubernetes_resource.infra-spot will be read during apply
  # (config will be reloaded to verify a check block)
 <= data "kubernetes_resource" "infra-spot" {
      + api_version = "apps/v1"
      + kind        = "Deployment"
      + object      = {
          + apiVersion = "apps/v1"
          + kind       = "Deployment"
          + metadata   = {
              + annotations                = {
                  + "deployment.kubernetes.io/revision" = "1"
                  + "meta.helm.sh/release-name"         = "karpenter"```

This will become unusable if we add sufficient checks to cover the estate… is there a way to suppress this output from the plan?

2

Aha I’ve realised a simple workaround here is to pull the data source out of the check block - it can still refer to it when externally defined, and now doesn’t pollute the plans.

3

Others may disagree, but I’d suggest if it’s at all possible to manage Kubernetes resources from within Terraform as little as possible (for example, using them to bootstrap a system like Flux or ArgoCD).

4

The fact that the data source is shown because it needs to be reloaded later is an artifact of how the renderer was designed before checks existed. I don’t see any reason the output needs to be shown, and the fact that you can put a data source within the check block is really the primary reason for their existence at all!

I’ll file an enhancement request to take a look at changing the output.