Unsupported format 'qcow2'

[flombois]

Hi Everybody

I am trying to automate the deployment of a Ubuntu cloud VM as shown in the provider plugin documentation: cloud-image, I know Proxmox does not natively supports qcow2 format but the documentation mention that the provider can do the conversion however I am getting the following error when applying the configuration:

terraform apply --auto-approve -var-file=.tfvars

2024-06-20T16:50:57.478+0200 [ERROR] vertex "proxmox_virtual_environment_vm.vault_vm" error: error waiting for VM creation: task "UPID:backup0:000E4EC2:18A7E4674:667441D0:qmcreate:102:root#@pam!terraform:" failed to complete with exit code: unable to create VM 102 - unsupported format 'qcow2' at /usr/share/perl5/PVE/Storage/LVMPlugin.pm line 366.
╷
│ Error: error waiting for VM creation: task "UPID:backup0:000E4EC2:18A7E4674:667441D0:qmcreate:102:root@pam!terraform:" failed to complete with exit code: unable to create VM 102 - unsupported format 'qcow2' at /usr/share/perl5/PVE/Storage/LVMPlugin.pm line 366.
│ 
│   with proxmox_virtual_environment_vm.vault_vm,
│   on resources.tf line 45, in resource "proxmox_virtual_environment_vm" "vault_vm":
│   45: resource "proxmox_virtual_environment_vm" "vault_vm" {
│ 
╵
2024-06-20T16:50:57.500+0200 [DEBUG] provider.stdio: received EOF, stopping recv loop: err="rpc error: code = Unavailable desc = error reading from server: EOF"
2024-06-20T16:50:57.500+0200 [DEBUG] provider: plugin process exited: path=.terraform/providers/registry.terraform.io/bpg/proxmox/0.60.0/darwin_arm64/terraform-provider-proxmox_v0.60.0 pid=30083
2024-06-20T16:50:57.501+0200 [DEBUG] provider: plugin exited

I am running Proxmox 7.2

This is my terraform deployment configurations:

resources.tf

# Ubuntu
locals {
  img_repository         = "https://cloud-images.ubuntu.com"
  img_os                 = "ubuntu"
  img_codename           = "noble"
  img_version            = "server"
  img_release            = "20240612"
  img_type               = "cloudimg"
  img_arch               = "amd64"
  img_format             = "img"
  img_checksum           = "784f6811c7681110cb066c8b4ad2a8fa654ad9c3ed63278286389f4180377600"
  img_checksum_algorithm = "sha256"
  img_filename           = "${local.img_codename}-${local.img_version}-${local.img_type}-${local.img_arch}.${local.img_format}"
  img_url                = "${local.img_repository}/${local.img_codename}/${local.img_release}/${local.img_filename}"
}

# Download the source image
resource "proxmox_virtual_environment_download_file" "img_source" {
  content_type       = "iso"
  datastore_id       = var.pve_node_img_storage_id
  file_name          = local.img_filename
  node_name          = var.pve_node
  url                = local.img_url
  checksum           = local.img_checksum
  checksum_algorithm = local.img_checksum_algorithm
}

# Create a proxmox vm and configure it through cloud init
resource "proxmox_virtual_environment_vm" "vault_vm" {
  depends_on  = [proxmox_virtual_environment_download_file.img_source]
  name        = "Vault"
  vm_id       = var.vm_id
  description = "Hashicorp vault instance managed by Terraform"
  tags        = ["terraform", "ubuntu", "hashicorp", "vault"]
  node_name   = var.pve_node

  startup {
    order = "3"
    #up_delay   = "60"
    #down_delay = "60"
  }

  agent {
    enabled = false
  }

  stop_on_destroy = true

  cpu {
    cores   = var.vm_cpu_cores
    sockets = var.vm_cpu_sockets
  }

  memory {
    dedicated = var.vm_memory
  }

  network_device {
    bridge = "vmbr1"
  }

  # System disk
  disk {
    interface    = "virtio0"
    datastore_id = var.vm_system_disk_datastore
    file_id      = proxmox_virtual_environment_download_file.img_source.id
    size         = var.vm_system_disk_size
  }

  # Data disk
  disk {
    interface    = "scsi0"
    datastore_id = var.vm_data_disk_datastore
    size         = var.vm_data_disk_size
  }

  initialization {
    ip_config {
      ipv4 {
        address = "dhcp"
      }
    }
  }

  operating_system {
    type = "l26"
  }
}

main.tf

terraform {
  required_providers {
    proxmox = {
      source  = "bpg/proxmox"
      version = "0.60.0"
    }
  }
}

locals {
  api_token = "${var.pve_username}!${var.pve_token_id}=${var.pve_token_secret}"
}

provider "proxmox" {
  endpoint  = var.pve_endpoint
  api_token = local.api_token
  insecure  = var.pve_insecure

  ssh {
    agent    = true
    username = var.pve_ssh_username
    password = var.pve_ssh_password
  }
}

If I try to use file_format = 'raw' I am getting a different error:

2024-06-20T16:52:56.757+0200 [ERROR] provider.terraform-provider-proxmox_v0.60.0: Response contains error diagnostic: tf_proto_version=6.6 tf_rpc=ApplyResourceChange tf_resource_type=proxmox_virtual_environment_vm @caller=github.com/hashicorp/[email protected]/tfprotov6/internal/diag/diagnostics.go:58 @module=sdk.proto diagnostic_detail="" diagnostic_severity=ERROR
  diagnostic_summary=
  | resizing disk: error waiting for VM disk resize: All attempts fail:
  | #1: the server did not include a data object in the response
   tf_provider_addr=registry.terraform.io/bpg/proxmox tf_req_id=2802ca8b-e2b7-ae12-5324-546c9c390e09 timestamp="2024-06-20T16:52:56.757+0200"
2024-06-20T16:52:56.780+0200 [DEBUG] State storage *statemgr.Filesystem declined to persist a state snapshot
2024-06-20T16:52:56.780+0200 [ERROR] vertex "proxmox_virtual_environment_vm.vault_vm" error: resizing disk: error waiting for VM disk resize: All attempts fail:
#1: the server did not include a data object in the response
╷
│ Error: resizing disk: error waiting for VM disk resize: All attempts fail:
│ #1: the server did not include a data object in the response
│ 
│   with proxmox_virtual_environment_vm.vault_vm,
│   on resources.tf line 45, in resource "proxmox_virtual_environment_vm" "vault_vm":
│   45: resource "proxmox_virtual_environment_vm" "vault_vm" {
│ 
╵
2024-06-20T16:52:56.791+0200 [DEBUG] provider.stdio: received EOF, stopping recv loop: err="rpc error: code = Unavailable desc = error reading from server: EOF"
2024-06-20T16:52:56.794+0200 [DEBUG] provider: plugin process exited: path=.terraform/providers/registry.terraform.io/bpg/proxmox/0.60.0/darwin_arm64/terraform-provider-proxmox_v0.60.0 pid=30096
2024-06-20T16:52:56.794+0200 [DEBUG] provider: plugin exited

Any clue how I can enable the image format conversion ?

[bpg]

Hi @flombois 👋🏼

It might be related to Proxmox 7.2 -- this version is not officially supported by the provider, and I'm not testing against it. But I haven't look deeper just yet.

There is also a similar issue in #1404

[flombois]

Hi @bpg , thanks for your quick answer

I'll try to upgrade Proxmox as soon as possible to check whether the same issue happens again.

[flombois]

I just upgraded my Proxmox from 7.2 -> 7.4 -> 8.2 as mentioned in https://pve.proxmox.com/wiki/Upgrade_from_7_to_8#Continuously_use_the_pve7to8_checklist_script using 'in-place' upgrade. I tried to redeploy with the exact same config however I am running in the same exact issue.

However if I use file_format = 'raw' then it runs fine now and create the disks. Maybe my Proxmox instance is not properly configured to support 'qcow2' format. I'll continue investigating.

[bpg]

The documentation for file_format is not super clear, and we can probably improve it.

This parameter defines not the source disk image file format, but the target format for the disk in the storage. And this depends on the storage type. Block storages, like LVM that you have, support only raw format, while directory-based storages allow you to select different formats, including qcow2.

Examples from the PVE UI:

Block storage:

Directory storage:

Hope this clarifies things a bit.

[flombois]

Thanks for those information, I'll check soon with different storage type and I think it should be the main issue indeed.

[1-cameron]

I'm having a similar issue with creating VMs with qcow2 disks on Proxmox VE 7.4 (I've had no issue on PVE 8) and 0.55.0 of the provider:

Error: resizing disk: error waiting for VM disk resize: All attempts fail

However, with version 0.46.1 this is not a problem so that can be a workaround for anyone who's stuck on PVE 7.4. Did something change between these versions that would affect disks in this way?

I'll try with the latest version of the provider soon to see if the issue persists.

[bpg]

Hi @1-cameron,

That error message was added in version 0.58.1 as part of #1336. The change involved using the PVE resize API instead of custom pvesh commands over SSH.
This API has changed in 8.x:

When resizing a disk, spawn a worker task to avoid HTTP request timeout (issue 2315).

As noted in the project's readme, the provider only supports the latest major version of PVE, so unfortunately, breakages with older versions will occur. I don't have the resources to test or maintain backward compatibility with PVE 7.x across the entire feature set.