Remove limit for taints in kubernetes_node_taint resource

[BBBmau]

Description

Fixes #2035

Removes the MaxItems line from the "taint" attribute. Should allow the ability of multiple taint blocks if desired.

Release Note

Release note for CHANGELOG:

`kubernetes/resource_kubernetes_node_taint.go`: Remove MaxItems from taint attribute

References

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

[arybolovlev]

This PR does not address the issue. Yes, Terraform is not throwing an error message anymore. However, the kubernetes_node_taint resource takes into account only the first taint in the list, the rest doesn't apply. As the result, we get diff with every plan run.

Another point here is that tests were not updated to validate changes. I would suggest here the TDD approach. The idea is you first write a test that reproduces an issue without changing the code. In short, the tests should fail at this point. Once you have that done, you make changes in the code and run tests again, if the issue is addressed, the tests will pass.

[BBBmau]

Tried the case where a user will add taints with the same key but different values. A panic is outputted but the tfstate still shows the applied config.

tfconfig:

provider "kubernetes" {
  config_path    = "~/.kube/config"
}

resource "kubernetes_node_taint" "tainttest" {
  metadata {
    name = "minikube"
  }

  taint {
    key    = "node-role.kubernetes.io/example"
    value  = "false"
    effect = "NoSchedule"
  }
    taint {
    key    = "node-role.kubernetes.io/example"
    value  = "true"
    effect = "NoSchedule"
  }
}

tfstate:

            "taint": [
              {
                "effect": "NoSchedule",
                "key": "node-role.kubernetes.io/example",
                "value": "false"
              },
              {
                "effect": "NoSchedule",
                "key": "node-role.kubernetes.io/example",
                "value": "true"
              }
            ]

doing terraform refresh turns it back to the state before, but this extra step shouldn't be needed.

[jrhouston]

@BBBmau This is expected behaviour, you can't set multiple taints with the same key. We should add some validation here so that it fails on the plan instead of generating the error at apply time.

[jrhouston]

The list type doesn't support validators so I've pushed some logic that does the validation in a custom diff.

[jrhouston]

Now it will show on plan time:

terraform plan
╷
│ Error: taint: duplicate taint key "node-role.kubernetes.io/example": taint keys must be unique
│
│   with kubernetes_node_taint.tainttest,
│   on main.tf line 5, in resource "kubernetes_node_taint" "tainttest":
│    5: resource "kubernetes_node_taint" "tainttest" {
│
╵

Review resolved