New livedns record: context deadline exceeded

[dan-mcdonald]

I'm trying to add a new A record to my Gandi LiveDNS with this provider.

resource "gandi_livedns_record" "agency" {
  name = "agency"
  ttl = 300
  type = "A"
  values = [aws_instance.agency.public_ip]
  zone = "▮▮▮▮▮.net"
}

But during apply I get this error (run with TF_LOG=INFO):

gandi_livedns_record.agency: Creating...
2023-07-05T10:44:10.239-0500 [INFO]  Starting apply for gandi_livedns_record.agency
2023-07-05T10:44:18.053-0500 [ERROR] vertex "gandi_livedns_record.agency" error: Fail to do the request (error 'Get "https://api.gandi.net/v5/livedns/domains/▮▮▮▮▮.net/records/agency/A": context deadline exceeded (Client.Timeout exceeded while awaiting headers)')
╷
│ Error: Fail to do the request (error 'Get "https://api.gandi.net/v5/livedns/domains/▮▮▮▮▮.net/records/agency/A": context deadline exceeded (Client.Timeout exceeded while awaiting headers)')
│
│   with gandi_livedns_record.agency,
│   on main.tf line 36, in resource "gandi_livedns_record" "agency":
│   36: resource "gandi_livedns_record" "agency" {
│

I think the request will work if it is given more time. My internet connection is not the fastest. Is there a configuration knob to increase the timeout? Currently it fails for me after only 10/15 seconds.

Other details:

> terraform version 
2023-07-05T10:48:33.039-0500 [INFO]  Terraform version: 1.5.2
2023-07-05T10:48:33.041-0500 [INFO]  Go runtime version: go1.20
2023-07-05T10:48:33.042-0500 [INFO]  CLI args: []string{"C:\\Users\\danie_10m7rom\\dev\\bin\\terraform.exe", "version"}
2023-07-05T10:48:33.062-0500 [INFO]  CLI command args: []string{"version"}
Terraform v1.5.2
on windows_amd64
+ provider registry.terraform.io/go-gandi/gandi v2.2.3
+ provider registry.terraform.io/hashicorp/aws v5.6.1

Any help you provide is appreciated. I know I'm not entitled to any free help.

[benbouillet]

I'm getting this issue intermittently, did you find a workaround ? @dan-mcdonald

[marcaurele]

We do have the same problem and it appears quite often on our plan execution, which is very problematic for us. I'm surprised to read from the API documentation^1:

Rate limit

Gandi's API is rate limited; you can perform a maximum of 1000 requests per minute from the same IP address.

which we are definitively not hitting.

Moreover, looking at the code / reference there is a (strange) custom default key to configure the timeout for the nemserver records: https://registry.terraform.io/providers/go-gandi/gandi/latest/docs/resources/nameservers#default

Adding this block in the resource did not even fix the problem:

resource "gandi_nameservers" "example" {
  ...
  timeouts {
    default = 60
  }
}

[!NOTE] The provider code has some code for a default timeout, which is not exposed in the provider configuration, with a default value of 5 seconds.

:shrug:

[insertjokehere]

I'm running into this essentially every time I run terraform plan, which is very frustrating.

It seems like the plugin passes config back to the go-gandi library here:

https://github.com/go-gandi/terraform-provider-gandi/blob/c38a15f10015d5d14a8982ca334709e241092a93/gandi/provider.go#L84-L106

The library exposes a timeout option (defaulting to 5 seconds) which the terraform provider doesn't provide a way to override:

https://github.com/go-gandi/go-gandi/blob/29df69368462adc734b4ffc650d33918f11d25f4/config/config.go#L6-L31

Ideally Gandi would fix their API so it wasn't so slow, but as an interim solution it'd be nice to be able to override the timeout config

[marcaurele]

We solved it the hard way... moving away from Gandi.