[Bug] HTTP 400 Bad Request When Creating elasticstack_kibana_security_role

[browol]

Describe the bug

I encountered an error while attempting to create the elasticstack_kibana_security_role resource. Although the Kibana role appears to be created successfully, the Terraform resource (elasticstack_kibana_security_role.role) fails to be created.

Upon reviewing the Kibana logs, it seems the issue is caused by a 400 Bad Request response. Specifically:

1. A PUT request to /api/security/role/developer_role?createOnly=true is successfully processed.
2. Immediately afterward, a GET request to /api/security/role/developer_role?createOnly=true is sent, which results in a 400 Bad Request error from the Kibana server.

This behavior suggests that the provider's code is making an unnecessary or improperly formatted GET request after the successful PUT.

I believe the problem is related to the implementation at the following code location:

https://github.com/elastic/terraform-provider-elasticstack/blob/v0.11.11/internal/kibana/role.go#L273

Could you please investigate and provide a solution?

To Reproduce
Steps to reproduce the behavior:

1. TF configuration used '...'

resource "elasticstack_kibana_security_role" "role" {
  name = "developer_role"

  elasticsearch {
    cluster = []
    indices {
      names      = ["application-dev-*"]
      privileges = ["read"]
    }
    run_as = []
  }

  kibana {
    feature {
      name       = "dashboard"
      privileges = ["minimal_read"]
    }

    feature {
      name       = "discover"
      privileges = ["minimal_read"]
    }

    spaces = ["default"]
  }
}

2. TF operations to execute to get the error '...'

tofu apply

3. See the error in the output '...'

2024-11-27T18:30:42.394+0700 [ERROR] provider.terraform-provider-elasticstack_v0.11.11: Response contains error diagnostic: @caller=github.com/hashicorp/terraform-plugin-go@v0.24.0/tfprotov6/internal/diag/diagnostics.go:58 @module=sdk.proto diagnostic_severity=ERROR diagnostic_summary="400 Bad Request" tf_resource_type=elasticstack_kibana_security_role diagnostic_detail="" tf_provider_addr=registry.terraform.io/elastic/elasticstack tf_rpc=ApplyResourceChange tf_proto_version=6.6 tf_req_id=da98b79c-16ca-15e9-6e5f-c3b209d16c3c timestamp="2024-11-27T18:30:42.394+0700"
2024-11-27T18:30:42.400+0700 [DEBUG] State storage *statemgr.Filesystem declined to persist a state snapshot
2024-11-27T18:30:42.400+0700 [ERROR] vertex "elasticstack_kibana_security_role.role" error: 400 Bad Request
╷
│ Error: 400 Bad Request
│ 
│   with elasticstack_kibana_security_role.role,
│   on main.tf line 72, in resource "elasticstack_kibana_security_role" "role":
│   72: resource "elasticstack_kibana_security_role" "role" {
│ 
╵
2024-11-27T18:30:42.408+0700 [DEBUG] provider.stdio: received EOF, stopping recv loop: err="rpc error: code = Unavailable desc = error reading from server: EOF"
2024-11-27T18:30:42.409+0700 [DEBUG] provider: plugin process exited: path=.terraform/providers/registry.opentofu.org/elastic/elasticstack/0.11.11/darwin_arm64/terraform-provider-elasticstack_v0.11.11 pid=74063
2024-11-27T18:30:42.409+0700 [DEBUG] provider: plugin exited

Expected behavior
Expected to able to create Kibana role without any errors.

Debug output
Run terraform command with TF_LOG=debug and provide extended information on TF operations. Please ensure you redact any base64 encoded credentials from your output.
eg

2024-11-27T18:30:42.394+0700 [ERROR] provider.terraform-provider-elasticstack_v0.11.11: Response contains error diagnostic: @caller=github.com/hashicorp/terraform-plugin-go@v0.24.0/tfprotov6/internal/diag/diagnostics.go:58 @module=sdk.proto diagnostic_severity=ERROR diagnostic_summary="400 Bad Request" tf_resource_type=elasticstack_kibana_security_role diagnostic_detail="" tf_provider_addr=registry.terraform.io/elastic/elasticstack tf_rpc=ApplyResourceChange tf_proto_version=6.6 tf_req_id=da98b79c-16ca-15e9-6e5f-c3b209d16c3c timestamp="2024-11-27T18:30:42.394+0700"
2024-11-27T18:30:42.400+0700 [DEBUG] State storage *statemgr.Filesystem declined to persist a state snapshot
2024-11-27T18:30:42.400+0700 [ERROR] vertex "elasticstack_kibana_security_role.role" error: 400 Bad Request
╷
│ Error: 400 Bad Request
│ 
│   with elasticstack_kibana_security_role.role,
│   on main.tf line 72, in resource "elasticstack_kibana_security_role" "role":
│   72: resource "elasticstack_kibana_security_role" "role" {
│ 
╵
2024-11-27T18:30:42.408+0700 [DEBUG] provider.stdio: received EOF, stopping recv loop: err="rpc error: code = Unavailable desc = error reading from server: EOF"
2024-11-27T18:30:42.409+0700 [DEBUG] provider: plugin process exited: path=.terraform/providers/registry.opentofu.org/elastic/elasticstack/0.11.11/darwin_arm64/terraform-provider-elasticstack_v0.11.11 pid=74063
2024-11-27T18:30:42.409+0700 [DEBUG] provider: plugin exited

Screenshots
If applicable, add screenshots to help explain your problem.

Versions (please complete the following information):

• OS: MacOS M1
• Opentofu Version: 1.6.2
• Provider version: 0.11.11
• Kibana Version: 8.16.0

Additional context

It appears that the Kibana Role API (v8) does not support the createOnly query parameter in the GET method.

https://www.elastic.co/docs/api/doc/kibana/v8/operation/operation-get-security-role-name

[lockness-Ko]

Seeing the same thing here, same symptoms, same error, with version 8.17.0 on docker compose (https://elastic.co/start-local) with provider version 0.11.11:

[jakubpetrovic]

Facing the same problem with new elastic cloud deployments.

Os: Mac 15.2 & Ubuntu 22.04
Deployment version: 8.17.0
Terraform version: 1.9.8
Provider version: 0.11.11 & 0.11.12

[jelmokri]

have same issue with Elastic 8.16.0 & provider: 0.11.6
terraform verison: 1.10.2

[biscout42]

I assume it will be fixed in the linked PR above.

[tobio]

@biscout42 are we able to close this issue out now?

[biscout42]

Yep, the fix for createOnly should be in linked PR. Folks, could you try again with the v0.11.13? In case of errors feel free to create a new issue.

[jakubpetrovic]

Seems to work fine now on 8.17, thanks for the fix.

[biscout42]

Thanks for verifying.