Cannot omit aio disk setting (Proxmox Default), causing unwanted drift and forced reboots on VM import

[wnsh7420]

When managing a proxmox_virtual_environment_vm resource, the aio attribute inside the disk block has a hardcoded default of "io_uring". This prevents users from omitting the aio setting entirely, which is required to match Proxmox's implicit Default (io_uring) state in the GUI (meaning the aio option is completely absent from the VM's .conf file).

Because the provider forces aio=io_uring in the API payload, it is impossible to cleanly import existing VMs created via the Proxmox Web GUI without introducing a configuration drift. Attempting to apply this drift forces an explicit write to the .conf file, which consequently triggers an unwanted VM reboot.

• bpg/proxmox v0.104.0

Affected Resource(s)

• proxmox_virtual_environment_vm

Expected Behavior

1. Users should be able to set aio = null (or omit the attribute) to signify that Terraform should not send the aio parameter in the API payload at all.
2. When importing an existing VM that has "Default (io_uring)" in the Proxmox GUI (no explicit aio in .conf), terraform plan should show no changes.

Actual Behavior
Even if aio is omitted from the HCL configuration, the provider injects aio = "io_uring".
When importing an existing VM that uses the implicit Proxmox default, terraform plan detects a drift:

~ aio = null -> "io_uring"

[bpg-dev]

Hey @wnsh7420 👋🏼
This is a known issue falling under umbrella of mishandled defaults in the legacy SDK resource.

Will be fixed when we migrate VM to the plugin framework in #1231. The behaviour is well understood and requirements are captured in ADR-004