Bug: timescale_peering_connection provisioned_id is null before refresh

[eddyfussel]

Hi, I'm using this provider to create a timescale database for testing the saas product. Since we would like to use a private endpoint, I use the timescale_peering_connection resource to peer with our vpc.

If I place my aws_vpc_peering_connection_accepter directly next to the timescale_peering_connection I can directly use the provisioned_id as an input for the vpc_peering_connection_id

Here is some example code

resource "timescale_peering_connection" "peering" {
  peer_account_id  = var.aws_account_id
  peer_region_code = var.aws_region
  peer_vpc_id      = var.vpc_id
  timescale_vpc_id = timescale_vpcs.vpc.id
}

resource "aws_vpc_peering_connection_accepter" "accepter_to_timescale" {
  vpc_peering_connection_id = timescale_peering_connection.peering.provisioned_id
  auto_accept               = true

  accepter {
    allow_remote_vpc_dns_resolution = true
  }

  tags = {
    Name        = "timescale/${timescale_vpcs.vpc.name}"
    description = "Peer ${var.aws_account_id}/${var.vpc_id} to timescale/${timescale_vpcs.vpc.name} for project_id ${timescale_vpcs.vpc.project_id}"
    side        = "accepter"
  }
}

However this is not working, as get following error that indicates, that the provisioned-id is null.

│ Error: Missing required argument
│
│   with aws_vpc_peering_connection_accepter.accepter_to_timescale,
│   on main.tf line 14, in resource "aws_vpc_peering_connection_accepter" "accepter_to_timescale":
│   14:   vpc_peering_connection_id = timescale_peering_connection.peering.provisioned_id
│
│ The argument "vpc_peering_connection_id" is required, but no definition was
│ found.

To proof that I directly pulled the state from terraform by executing terraform console

❯ terragunt console
zsh: command not found: terragunt
❯ terragrunt console
INFO[0000] Terragrunt Cache server is listening on 127.0.0.1:50798
INFO[0000] Start Terragrunt Cache server
> timescale_peering_connection.peering
{
  "error_message" = tostring(null)
  "id" = tonumber(null)
  "peer_account_id" = "0123456789"
  "peer_cidr" = tostring(null)
  "peer_region_code" = "eu-central-1"
  "peer_vpc_id" = "vpc-012345467890"
  "provisioned_id" = tostring(null)
  "status" = tostring(null)
  "timescale_vpc_id" = 1234
  "vpc_id" = tostring(null)
}
> exit

Executing the apply again I get a change on the resource and the code starts working.

  # timescale_peering_connection.peering has changed
  ~ resource "timescale_peering_connection" "peering" {
      + provisioned_id   = "pcx-0123456789"
        # (4 unchanged attributes hidden)
    }

I would expect this to work, as the resource aws_vpc_peering_connection from aws to create a vpc peering between multiple vpcs is waiting until there is the pcx-.. available.

I've tried to update my code to add a delay, unfortunately that wasn't successful.

resource "timescale_peering_connection" "peering" {
  peer_account_id  = var.aws_account_id
  peer_region_code = var.aws_region
  peer_vpc_id      = var.vpc_id
  timescale_vpc_id = timescale_vpcs.vpc.id

  # We delay here the creation of the peering, as the provisioned_id is null until aws is ready
  # and unfortunatly the provisioner directly gives us this null object back without waiting for
  # aws to be ready. Otherwise we can't use this in aws_vpc_peering_connection_accepter object.
  provisioner "local-exec" {
    command = "sleep 90"
  }
}

I also have here a log file to see the commands I've executed:
bug_timescale_provider_retracted.txt

Thanks in advance for any feedback, idea or help.

BR eddyfussel

[Khyme]

Hello @eddyfussel,
I encountered the same issue myself when developing it, and had some issues fixing it.
IIRC, the code handling the timescale peering connection doesn't surface the provisioned-id so I think it needs wider code change on our end I didn't have the opportunity to, so released that half way done.
I'm sorry for the inconvenience.

As for when this might be fixed, I think this provider will not see some love again until Q3/4.