IAM Role creation race stil present in spot instance requests

[ghost]

This issue was originally opened by @stadeltobi as hashicorp/terraform#17457. It was migrated here as a result of the provider split. The original body of the issue is below.

---

Terraform Version

Terraform v0.10.2

resource "aws_spot_instance_request" "quick-sir" {
  ami = "ami-5652ce39"
  instance_type = "t2.xlarge"
  spot_price = ".15"
  wait_for_fulfillment = true
  vpc_security_group_ids = <...>
  key_name = <...>
  iam_instance_profile = "${aws_iam_instance_profile.foo.id}"
  user_data = "${data.template_file.cloud-init.rendered}"
}

...

Additional Context

I'm having the equivalent issue that is described in hashicorp/terraform#15341 when, instead of trying to create an aws_instance, I use a aws_spot_instance_request.

The same error ("Invalid Parameter Value") is thrown, so I'm assuming that the race condition described above was fixed for regular (on-demand) instances still prevails for spot_instance_requests.

Also, the workaround suggested by @spanktar makes the problem go away.

The problem goes away if I switch the spot request resource with an aws_instance resource, using the exact same parameters (except for the ones like spot_price that don't apply for aws_instance

References

• Invalid IAM Instance Profile name terraform#15341

[bflad]

@stadeltobi can you confirm which version of the AWS provider you are running? It does look like the aws_spot_instance_request resource of current provider (v1.10.0) only has 15 seconds of retry where we normally retry for a minute or two when dealing with IAM eventual consistency.

[stadeltobi]

@bflad thanks for getting back to me!

It's version 1.10.0 of the aws provider.

And you're probably right: in my setup, the error is thrown exactly 15 seconds after attempting to create the respective aws_spot_instance_request ...

[bflad]

Nice - let's bump that up to a minute to match other retry logic 😄 Would you feel comfortable submitting a PR? No acceptance test changes required.

https://github.com/terraform-providers/terraform-provider-aws/blob/fefe3470fe9aadca6e84a79bcc56b46a3f420952/aws/resource_aws_spot_instance_request.go#L139

[stadeltobi]

sure, why not :)

#3561

[bflad]

PR above has been merged and will release in v1.11.0 of the AWS provider, likely releasing next week. Of course you're more than welcome to try out a custom local build too. 😉

[stadeltobi]

I'll do that; thanks a lot for the quick response and the help.

[stadeltobi]

So now I did (built c56c678, so a few commits after the PR was merged).

Works like a charm.

Thanks again 😄

[bflad]

This has been released in version 1.11.0 of the AWS provider. Please see the Terraform documentation on provider versioning or reach out if you need any assistance upgrading.

[asgoel]

Hi all, we recently started seeing this happening again, on provider version 1.42. Given that we haven't updated the provider recently, could this be due to some slowness on AWS' side?

[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 feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!