`HTTP 500` on some resources when reinstalling `Proxmox` about terraform-provider-proxmox HOT 4 OPEN

Hi @aleprovencio! 👋🏼

Was there any error messages reported by PVE? These issues are quite hard to debug without reproducing, which means quite a bit of efforts with reinstalling PVE. So any additional details is really appreciated.

In general, if you reset the remote state of the resource (i.e. deleted the resource outside of terraform), the local TF state should be also deleted, so there is no inconsistency or "state drift" for the provider to reconcile.

from terraform-provider-proxmox.

Hello @bpg, thanks for the reply and of course, also for this awesome project.

It does makes sense to me that I should probably remove all proxmox related resources from terraform state prior reinstalling it, in order to prevent the so called "state drift".

However I still would like to understand why resources like proxmox_virtual_environment_vm or proxmox_virtual_environment_container get recreated while resources mentioned on this issue do not, on a test like I did without manual interventions on the terraform state.

Regarding errors, I don't see anything special on PVE's side and on terraform's, besides the HTTP 500 it just says that those resources do not exist. I wish I could give you additional details on the problem, but that's all I have for now, maybe I could try better debugging with your help.

from terraform-provider-proxmox.

It looks like the affected resource are "compound resources", i.e. they have references to other separate proxmox entities that are on different API paths. When provider applies a change, first it has to read the resource state from the remote to detect the "drift". I think there are logical or implementation bugs in those resources,they probably are trying to read the dependent objects first (like ACLs for a user, or rules for a security group) using the "parent" object ID as a request criteria. Those parents do not exist, and requests fail.

That's my hypothesis, without any actual debugging. There is definitely something in the provider's implementation that can be improved in this regard, though a proper investigation is needed to make a fix.

from terraform-provider-proxmox.

Yeah I guess you are on the right path.

I've done a new test where I did remove those resources from state, reinstalled proxmox and although terraform apply seemed worked flawlessly the first time, issuing the same command again still suggested changes on these resources we talk about.

from terraform-provider-proxmox.