Adding tolerations causes plan failure with error `AttributeName("tolerationSeconds"): step cannot be applied to this value`

[soulshake]

Terraform, Provider, Kubernetes versions

Terraform version: v0.15.5
Provider version: v0.5.0
Kubernetes version: v1.21.1 (client), v1.20.4-eks-6b7464 (server)

Affected Resource(s)

• kubernetes_manifest

Terraform Configuration Files

resource "kubernetes_manifest" "runner" {                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      
  provider = kubernetes-alpha
 
  manifest = {
    apiVersion = "apps/v1"
    kind       = "Deployment"
    metadata = {
      name      = "test"
      namespace = "default"
    }
    spec = {
      selector = {
        matchLabels = {
          app = "example"
        }
      }
 
      template = {
        metadata = {
          labels = {
            app = "example"
          }
        }
 
        spec = {
          containers = [
            {
              image   = "alpine:latest"
              name    = "ping"
              command = ["sh", "-c"]
              args    = ["ping goo.gl"]
            }
          ]
          # causes planning to fail if added after the resource exists
          tolerations = [
            {
              effect   = "NoSchedule"
              key      = "nvidia.com/gpu"
              operator = "Exists"
            },
          ]
        }
      }
    }
  }
}

Provider config

terraform {
  required_providers {
    aws = {
      source  = "hashicorp/aws"
      version = "~> 3.46.0"
    }
    kubernetes-alpha = {
      source  = "hashicorp/kubernetes-alpha"
      version = ">= 0.5.0"
    }
  }
}
 
provider "aws" {
  region = "us-east-2"
}
 
provider "kubernetes-alpha" {
  host                   = data.aws_eks_cluster.target.endpoint
  cluster_ca_certificate = base64decode(data.aws_eks_cluster.target.certificate_authority.0.data)
  token                  = data.aws_eks_cluster_auth.target.token
}
 
data "aws_eks_cluster" "target" {
  name = "ci-apps"
}
 
data "aws_eks_cluster_auth" "target" {
  name = data.aws_eks_cluster.target.name
}

Debug Output

https://gist.github.com/soulshake/180f8264a7d89c70d929cf05585b8281

Panic Output

Steps to Reproduce

# first apply without tolerations
terraform apply
# then add tolerations and try to plan
terraform plan

Expected Behavior

The tolerations should be successfully planned/added.

Actual Behavior

Planning and applying when the resource has not yet been created is successful.

However, trying to add tolerations on an existing resource fails with an error: AttributeName("tolerationSeconds"): step cannot be applied to this value:

$ terraform plan
kubernetes_manifest.runner: Refreshing state...
╷
│ Error: Failed to update proposed state from prior state
│ 
│   with kubernetes_manifest.runner,
│   on main.tf line 32, in resource "kubernetes_manifest" "runner":
│   32: resource "kubernetes_manifest" "runner" {
│ 
│ AttributeName("spec").AttributeName("template").AttributeName("spec").AttributeName("tolerations").ElementKeyInt(0).AttributeName("tolerationSeconds"): step cannot be applied to this value

Observations:

• If I delete the deployment, I'm able to plan and apply successfully.

Important Factoids

n/a

References

n/a

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment