Scaleway: Adding multiple volumes doesn't work

[tboerger]

Terraform Version

0.7.6

Affected Resource(s)

• scaleway_server
• scaleway_volume
• scaleway_volume_attachment

Terraform Configuration Files

variable "access" {
  type = "string"
}

variable "org" {
  type = "string"
}

provider "scaleway" {
  access_key   = "${var.access}"
  organization = "${var.org}"
}

resource "scaleway_server" "store" {
  name  = "store"
  image = "75c28f52-6c64-40fc-bb31-f53ca9d02de9"
  type  = "C2S"
}

resource "scaleway_volume" "store" {
  count      = "3"
  name       = "vol-${count.index}"
  size_in_gb = "150"
  type       = "l_ssd"
}

resource "scaleway_volume_attachment" "store" {
  count  = "3"
  volume = "${element(scaleway_volume.store.*.id, count.index)}"
  server = "${element(scaleway_server.store.*.id, 0)}"
}

Debug Output

https://gist.github.com/tboerger/c1e36a1ab2d53cfc072c6a9919bcbc66

Panic Output

n/a

Expected Behavior

I would expect to have a running server that got 3 additional volumes assigned.

Actual Behavior

The setup simply fails because one volume attachment invokes a reboot of the server while the next volume attachment fails because the server is not running anymore:

2 error(s) occurred:

* scaleway_volume_attachment.store.2: StatusCode: 400, Type: invalid_request_error, APIMessage: server is being stopped or rebooted
* scaleway_volume_attachment.store.0: StatusCode: 400, Type: invalid_request_error, APIMessage: server is being stopped or rebooted

After the failed action you can see on the Scaleway web UI that only one volume have been attached to the server:

### Steps to Reproduce 1. Get an `access_key` and the `organization` from your Scaleway account 2. Simply execute `terraform apply` and wait ### Important Factoids

I tried to launch multiple nodes with multiple volumes at once, but I have slimmed down the example to keep it simple. Maybe it would already help to assign volumes while node creation, the API supports that.

References

Maybe these issues are somehow related or maybe a PR solves the issue partially:

• scaleway instance with not default volumes #9254
• Improvements for Scaleway machine creation #8678

/cc @nicolai86

[nicolai86]

Hey @tboerger

thanks for the bug report!

I think the server state handling inside scaleway_volume_attachment is the problem.
Scaleway only allows to attach volumes to a stopped server, meaning the resource needs to stop the server, attach the volume, and start it up again.

As a test for this hypothesis you could add state = "stopped" to the scaleway_server resource. If it works with a stopped server we'll have to make the state handling more robust.

Can you verify that?

[tboerger]

I thought that can really help, but that results in an server error: https://gist.github.com/tboerger/ec91591d98335f69187ba89cbdb49eb2

--- test.tf.orig    2016-10-18 07:51:34.000000000 +0200
+++ test.tf 2016-10-18 07:43:51.000000000 +0200
@@ -17,6 +17,7 @@
   name  = "store"
   image = "75c28f52-6c64-40fc-bb31-f53ca9d02de9"
   type  = "C2S"
+  state = "stopped"
 }

 resource "scaleway_volume" "store" {

[tboerger]

Maybe a solution can be to attach volumes directly to the server, the API accepts that (at least for creation). That can also solve #9254 and seems to be added within #8678, but than there is still the issue to add volumes to already created servers.

[nicolai86]

I think supporting volumes inline would work, and to work around the VC1M issue you referenced it's probably required.

The multiple volume attachment should be fixed in any case. Luckily adding some testcases should break this reliably, and then we can fix it.

[tboerger]

I really would like to provide a fix but I don't have enough internal knowledge of terraform to get an idea how to solve that issue with attaching volumes.

[nicolai86]

@tboerger not a nice solution, but if you're interested in the details check out #9493

[tboerger]

Getting solved via #9493

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.