Error when creating a compute instance

[wasadigi]

I am trying to provision a compute instance, but when I terraform apply, I get the following error:

Error: Post "https://api-ch-gva-2.exoscale.com/v2/instance": invalid request: bad request

Here is my config. I am not sure if I should specify a endpoint, as I have seen some examples with, and others without.

terraform {
  required_providers {
    exoscale = {
      source  = "exoscale/exoscale"
      version = ">=0.38.0"
    }
  }
}

provider "exoscale" {
  key = "${var.exoscale_api_key}"
  secret = "${var.exoscale_api_secret}"
  // compute_endpoint = "https://api.exoscale.com/compute"
}

locals {
  zone = "ch-gva-2"
}

data "exoscale_compute_template" "ubuntu" {
  zone = local.zone
  name = "Linux Ubuntu 22.04 LTS 64-bit"
}

resource "exoscale_compute_instance" "testing"  {
  zone = local.zone
  name = "test_name"
  type = "standard.medium"
  template_id = data.exoscale_compute_template.ubuntu.id
  disk_size = 10
  ssh_key = "olivier"
}

Note that the template is resolved correctly. I looked at the doc and I didn't see any required field for which I am not providing a value.

  # exoscale_compute_instance.testing will be created
  + resource "exoscale_compute_instance" "testing" {
      + created_at          = (known after apply)
      + disk_size           = 10
      + id                  = (known after apply)
      + ipv6                = false
      + ipv6_address        = (known after apply)
      + name                = "test_name"
      + private_network_ids = (known after apply)
      + public_ip_address   = (known after apply)
      + ssh_key             = "olivier"
      + state               = (known after apply)
      + template_id         = "54d8e676-d46a-4559-87e7-e0feed79e770"
      + type                = "standard.medium"
      + zone                = "ch-gva-2"
    }

[wasadigi]

I tried the example from https://registry.terraform.io/providers/exoscale/exoscale/latest/docs/resources/compute_instance. In this case, it creates the security group and the private network, but fails on the instance. The error message is different:

Error: Post "https://api-ch-gva-2.exoscale.com/v2/instance": resource not found

locals {
  zone = "ch-gva-2"
}

data "exoscale_compute_template" "ubuntu" {
  zone = local.zone
  name = "Linux Ubuntu 20.04 LTS 64-bit"
}

resource "exoscale_security_group" "web" {
  name = "web"
}

data "exoscale_security_group" "default" {
  name = "default"
}

resource "exoscale_private_network" "example" {
  name = "privnet"
  zone = local.zone

  start_ip = "10.0.0.20"
  end_ip   = "10.0.0.253"
  netmask  = "255.255.255.0"
}

resource "exoscale_compute_instance" "example" {
  zone               = local.zone
  name               = "webserver"
  type               = "standard.medium"
  template_id        = data.exoscale_compute_template.ubuntu.id
  disk_size          = 10
  security_group_ids = [
    data.exoscale_security_group.default.id, 
    exoscale_security_group.web.id,
  ]
  ssh_key            = "my-key"
  user_data          = <<EOF
#cloud-config
manage_etc_hosts: localhost
EOF

  network_interface {
    network_id = exoscale_private_network.example.id
    ip_address = "10.0.0.20"
  }
}

[wasadigi]

I thought that the missing security_group_ids might be the cause (although it is not described as required). So I updated my spec... but without success. Still the same bad request error. I have tried to trace terraform, but I don't see communication between the provider and the api.

data "exoscale_security_group" "default" {
  name = "default"
}

resource "exoscale_compute_instance" "testing"  {
  zone = local.zone
  name = "test_name"
  type = "standard.medium"
  template_id = data.exoscale_compute_template.ubuntu.id
  security_group_ids = [
    data.exoscale_security_group.default.id, 
  ]
  disk_size = 10
  ssh_key = "olivier"
}

[wasadigi]

Hum... using type = "gpu.small" gives me another error: "Post "https://api-ch-gva-2.exoscale.com/v2/instance": invalid request: This instance type is not allowed for your account".

I had the idea to try this, because I was looking for the list of accepted values for type. Is it available anywhere?

[wasadigi]

exo compute instance-type list seems to be what I was looking for... but using these values does not fix my original problem.

$ exo compute instance-type list
┼──────────────────────────────────────┼──────────┼─────────────┼
│                  ID                  │  FAMILY  │    SIZE     │
┼──────────────────────────────────────┼──────────┼─────────────┼
│ 71004023-bb72-4a97-b1e9-bc66dfce9470 │ standard │ micro       │
│ b6cd1ff5-3a2f-4e9d-a4d1-8988c1191fe8 │ standard │ tiny        │
│ 21624abb-764e-4def-81d7-9fc54b5957fb │ standard │ small       │
│ b6e9d1e8-89fc-4db3-aaa4-9b4c5b1d0844 │ standard │ medium      │
│ c6f99499-7f59-4138-9427-a09db13af2bc │ standard │ large       │
│ 350dc5ea-fe6d-42ba-b6c0-efb8b75617ad │ standard │ extra-large │
│ a216b0d1-370f-4e21-a0eb-3dfc6302b564 │ standard │ huge        │
│ c0d3fb5d-6fdb-4a63-9361-3e5cfa8b36d0 │ standard │ mega        │
│ 74bfaf4e-7d67-4adf-9322-12b9a36e84f7 │ standard │ titan       │
│ 07246b95-bbe4-498f-92a3-ce5be4d38342 │ gpu      │ small       │
│ dee65287-12cf-4e36-b635-32dbc9a2e909 │ gpu      │ huge        │
│ 011a5079-845d-467b-a660-8685dad1cb67 │ standard │ jumbo       │
│ 847b3a94-659f-445e-ba7e-d02f89355e1b │ gpu      │ medium      │
│ 1ee13a8a-38ec-4858-a7c5-82f707810ea3 │ gpu      │ large       │
│ f17df7cf-aced-4314-8709-4a0041408fdf │ storage  │ extra-large │
│ ebc4743c-9c7b-490d-8958-17a89561d9a3 │ storage  │ huge        │
│ a3bd08e2-bcdd-45b3-b6de-507d15056dae │ storage  │ mega        │
│ 964fd954-40ca-420f-871d-0aa7c5fac230 │ storage  │ titan       │
│ dec751cc-6e3b-4ac1-8a14-fdfdf46fd1f7 │ storage  │ jumbo       │
│ d141049f-71a3-4b1a-9016-3df0ac9a8cc9 │ gpu2     │ small       │
│ 98d53f21-06e2-4e38-8da7-d5e9ae570c83 │ gpu2     │ medium      │
│ 98981aba-d258-485b-89b2-52708768e88d │ gpu2     │ large       │
│ e61976b7-3527-4898-9b2c-382d7eeb49a0 │ gpu2     │ huge        │
│ 221d39c3-b2b9-4014-b32b-5d262f684eff │ cpu      │ extra-large │
│ 68c91810-55ed-4685-9771-00d9adf7b7c0 │ cpu      │ huge        │
│ c2f204df-29fc-4c20-b5c4-bdb21282cecc │ cpu      │ mega        │
│ e1ae0088-27fd-4185-a057-075ea0b0aa02 │ cpu      │ titan       │
│ dd6abc94-337f-4732-baff-ed745f8bd4f5 │ memory   │ extra-large │
│ efbf55db-01c9-4c73-973f-cecc4d26dcde │ memory   │ huge        │
│ 306eeeae-38e9-4add-897b-4eff0b019a6d │ memory   │ mega        │
│ 68a4780f-cc91-4df6-a735-d679318289bc │ memory   │ titan       │
│ 8a52e80d-b1d0-4efb-8059-a931fa33c5a1 │ standard │ colossus    │
│ 1552f778-726f-477f-85d2-92c3ca42e499 │ gpu3     │ small       │
│ 30d08069-5101-4029-a847-c5be5d916eb0 │ gpu3     │ medium      │
│ 2bf11dfd-5915-496d-bc0d-9f836f2fe7be │ gpu3     │ large       │
│ 5a426186-9f55-43a1-8e81-6736b2e4a73f │ gpu3     │ huge        │
┼──────────────────────────────────────┼──────────┼─────────────┼

[wasadigi]

Looks like the _ underscore in the name was the culprit...

[wasadigi]

Indeed:

[wasadigi]

The issue is solved. I keep it open, to ask one last question: to debug this type of problem, is there any way to get some trace to understand why the server is unhappy?

[kobajagi]

Hey @wasadigi , you can try running terraform commands with TF_LOG=DEBUG, that will also print full exoscale API responses.