[CCE] cce_node_pool_v3 import error

[mecofis]

Terraform provider version

Terraform v1.13.5
on linux_amd64

• provider registry.terraform.io/hashicorp/kubernetes v2.38.0
• provider registry.terraform.io/hashicorp/random v3.7.2
• provider registry.terraform.io/hashicorp/vault v5.4.0
• provider registry.terraform.io/opentelekomcloud/opentelekomcloud v1.36.52

Affected Resource(s)

opentelekomcloud_cce_node_pool_v3

Terraform Configuration Files

resource "opentelekomcloud_cce_node_pool_v3" "node_pool" {
  for_each = var.node_pools

  cluster_id = opentelekomcloud_cce_cluster_v3.this.id

  name                     = each.value.name
  flavor                   = each.value.flavor
  os                       = each.value.os
  availability_zone        = each.value.availability_zone
  initial_node_count       = each.value.initial_node_count
  min_node_count           = each.value.min_node_count
  max_node_count           = each.value.max_node_count
  scale_enable             = true
  scale_down_cooldown_time = 30
  priority                 = 1
  key_pair                 = each.value.key_pair

  root_volume {
    size       = each.value.root_volume_size
    volumetype = each.value.root_volume_type
    kms_id     = opentelekomcloud_kms_key_v1.key.id
  }

  data_volumes {
    size       = each.value.data_volume_size
    volumetype = each.value.data_volume_type
    kms_id     = opentelekomcloud_kms_key_v1.key.id
  }
}

Debug Output/Panic Output

terraform state show module.cluster.opentelekomcloud_cce_node_pool_v3.node_pool["az01"]

module.cluster.opentelekomcloud_cce_node_pool_v3.node_pool["az01"]:

resource "opentelekomcloud_cce_node_pool_v3" "node_pool" {
availability_zone = null
cluster_id = "dbb9d114-a298-11f0-96d6-02XXXXXXX40"
flavor = null
id = "ca0c13eb-a299-11f0-ac39-02XXXXXX03f"
k8s_tags = {}
key_pair = "id_rsa"
max_pods = 110
name = "ambersearch-cce-01-nodepool"
os = "HCE OS 2.0"
postinstall = null
preinstall = null
runtime = "containerd"
scale_enable = false
security_group_ids = []
status = null
subnet_id = "df2daefb-ba8b-4458-ad10-d64XXXXXXe9f"

data_volumes {
    extend_param  = null
    extend_params = {}
    kms_id        = "d88e21c5-f151-4438-b9ba-9aeXXXXXXX947"
    size          = 100
    volumetype    = "SSD"
}

root_volume {
    extend_param  = null
    extend_params = {}
    kms_id        = "d88e21c5-f151-4438-b9ba-9aedXXXXXX947"
    size          = 80
    volumetype    = "SSD"
}

timeouts {}

}

Steps to Reproduce

terraform import -var-file="envs/prod.tfvars" module.cluster.opentelekomcloud_cce_node_pool_v3.node_pool["az01"] CLUSTER_UUID/NODE_POOL_UUID

Expected Behavior

I thought after an import terraform has all necessary values.

Actual Behavior

After Import the node_pool i get no values for availability_zone and flavor. So if i do a plan terraform will create a new node pool.

Important Factoids

None

References

See API documentation.

[artem-lifshits]

Hello @mecofis I've tested import for cce nodepool.
availability_zone and flavor parameters are properly set in tfstate.

Are you sure that variables for flavor and availability_zone are provided properly? If not - it would cause an instance recreation during next apply.

[mecofis]

Hello @artem-lifshits
I dont see the parameters in my tfstate.
The variable are set in my tfvars file.
Yes my terraform want to recreate the nodepool

module.cluster.opentelekomcloud_cce_node_pool_v3.node_pool["az01"] must be replaced
-/+ resource "opentelekomcloud_cce_node_pool_v3" "node_pool" {
      + availability_zone        = "eu-de-01" # forces replacement
      + flavor                   = "m4.4xlarge.8" # forces replacement
      ~ id                       = "ca0c13eb-a299-11f0-ac39-0255XXXXX03f" -> (known after apply)
      + initial_node_count       = 3
      ~ k8s_tags                 = {} -> (known after apply)
      + max_node_count           = 3
      ~ max_pods                 = 110 -> (known after apply)
      + min_node_count           = 3
        name                     = "ambersearch-cce-01-nodepool"
      + postinstall              = (known after apply)
      + preinstall               = (known after apply)
      + priority                 = 1
      ~ runtime                  = "containerd" -> (known after apply)
      + scale_down_cooldown_time = 30
      ~ scale_enable             = false -> true
      ~ security_group_ids       = [] -> (known after apply)
      + status                   = (known after apply)
      ~ subnet_id                = "df2daefb-ba8b-4458-ad10-d6XXXXXf" -> (known after apply)
        # (3 unchanged attributes hidden)

      ~ data_volumes {
          - extend_params = {} -> null
            # (4 unchanged attributes hidden)
        }

      ~ root_volume {
          - extend_params = {} -> null
            # (4 unchanged attributes hidden)
        }

      - timeouts {}
    }

  "module": "module.cluster",
  "mode": "managed",
  "type": "opentelekomcloud_cce_node_pool_v3",
  "name": "node_pool",
  "provider": "provider[\"registry.terraform.io/opentelekomcloud/opentelekomcloud\"]",
  "instances": [
    {
      "index_key": "az01",
      "schema_version": 0,
      "attributes": {
        "agency_name": null,
        "availability_zone": "",
        "cluster_id": "dbb9d114-a298-11f0-96d6-025XXXXXXXX40",
        "data_volumes": [
          {
            "extend_param": "",
            "extend_params": {},
            "kms_id": "d88e21c5-f151-4438-b9ba-9aXXXXXX947",
            "size": 100,
            "volumetype": "SSD"
          }
        ],
        "docker_base_size": null,
        "docker_lvm_config_override": null,
        "flavor": "",

......

[artem-lifshits]

@mecofis could you provide TF debug output while importing the node pool?

You will need to add env vars before running import again:

export TF_LOG=DEBUG
export TF_LOG_PATH=./terraform.log

[mecofis]

@artem-lifshits here is the log file
this is my import command:
terraform import -var-file="envs/prod.tfvars" module.cluster.opentelekomcloud_cce_node_pool_v3.node_pool["az01"] dbb9d114-a298-11f0-96d6-02550a100040/ca0c13eb-a299-11f0-ac39-02550a10003f

terraform.log

[artem-lifshits]

@mecofis thank you for providing the log, but it doesn't contain the necessary debug information about the node pool v3 import. Could you show the import debug output in a separate resource without using modules?

[mecofis]

@artem-lifshits

terraform import opentelekomcloud_cce_node_pool_v3.node_pool dbb9d114-a298-11f0-96d6-02550a100040/ca0c13eb-a299-11f0-ac39-02550a10003f

I get the same error no flavor and no availability_zone

terraform.log

resource "opentelekomcloud_cce_node_pool_v3" "node_pool" {

  cluster_id = "dbb9d114-a298-11f0-96d6-02550a100040"

  name                     = "ambersearch-cce-01-nodepool"
  flavor                   = "m4.4xlarge.8"
  os                       = "HCE OS 2.0"
  availability_zone        = "eu-de-01"
  initial_node_count       = 1
  min_node_count           = 4
  max_node_count           = 3
  scale_enable             = true
  scale_down_cooldown_time = 30
  priority                 = 1
  key_pair                 = "id_rsa"

  root_volume {
    size       = 80
    volumetype = "SSD"
    kms_id     = "d88e21c5-f151-4438-b9ba-XXXXXXX947"
  }

  data_volumes {
    size       = 100
    volumetype = "SSD"
    kms_id     = "d88e21c5-f151-4438-b9ba-XXXXXXX947"
  }
}

provider "opentelekomcloud" {
  tenant_name = "eu-de"
  auth_url    = "https://iam.eu-de.otc.t-systems.com/v3"
}

terraform {
  required_version = "~>1"
  required_providers {
    opentelekomcloud = {
      source  = "opentelekomcloud/opentelekomcloud"
      version = "~>1.36"
    }
  }
}

[dombisza]

If you are still using for_each to manage multiple nodepools with a single resource the import command should be something like this:
terraform import "module.cce.opentelekomcloud_cce_node_pool_v3.this[\"<for_each_KEY>\"]" "${CLUSTER_ID}/${NODEPOOL_ID}"

[mecofis]

In the latest example there is no for_each.

[artem-lifshits]

@mecofis tried reproducing this issue with same TF version (1.13.5) and S3 credentials but it still works fine for me.
Could you also provide your cluster settings?

[mecofis]

Hello this is our node_pool specs from prod and dev:

prod.log
dev.log

.....
			"flavor": "m4.2xlarge.16",
			"az": "eu-de-01",
.....

I can see there differents in the nodeTemplate. In dev you can see az and flavor.
We did some changes in prod in the OTC console manually by switching the flavor.

I seems that this change caused our problem.