How to Handle AWS Spot Instance args (tags, etc.)

[johnnyshields]

This is a dupe of issue #3263 and PR #4380. I am raising a new issue because I'd like to revisit the problem and agree on an approach with the Terraform team.

The Problem

• We need to set tags on AWS spot instances.
• The Terraform schema contains a spot_instance_request resource. A aws_spot_instance_request is not an "instance" itself, rather it is a request to Amazon to generate an instance.
• The Terraform schema allows setting tags on aws_spot_instance_request, however, the tags are not forwarded by AWS to the actual instance when the request in fulfilled (see: http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-requests.html)

Possible Solutions

1. Interpret tags on aws_spot_instance_request to mean the tags to set on the instance created when the spot request is fulfilled. This is what PR provider/aws: tag the spot instance. #4380 does.
2. Add an instance_args node to the aws_spot_instance_request resource which allows various args like tags to be set by Terraform on the instance after fulfillment.
3. Enhance the aws_instance to have new fields is_spot_instance, spot_price, spot_type etc. If these are set, then Terraform uses a spot instance request workflow to initialize the instance. This approach will require the most effort but is ultimately the most friendly / least surprising to the developer in my opinion--many AWS-related libraries use this abstraction pattern.
4. Introduce a new resource aws_spot_instance which behaves as described in config: require blocks for upstream dependencies #3, to avoid changing the aws_instance object.

[johnnyshields]

According to the docs:

Spot Instance Requests support all the same arguments as aws_instance

and most of these args (security_groups, etc.) except tags seemed to be passed thru to the created instance by AWS itself.

[johnnyshields]

One more comment, to meet the immediate need I'd be in favor of merging PR #4380 even if a different option is chosen in the future (they aren't mutually exclusive.)

[cbarbour]

This is the documented behavior of AWS spot instances, rather than a bug in the AWS provider.

As a workaround, consider using an auto-scaling group. The autoscaling group can request spot instances, and the launch_configuration will pass tags onto the instance. You don't need to autoscale the autoscaling group; you can specify the pool size manually.

The downside to that approach, is that the autoscaling resources do not provide attributes for the resulting instances.

[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.