BREAKING CHANGE: update module to include latest features and remove use of count for module count/for_each

[bryantbiggs]

Description

• update module to include latest features and remove use of count for module count/for_each
• basic example renamed to complete - no longer just basic and shows multiple examples for more of a complete setup

New attributes added

• hibernation
• secondary_private_ips
• capacity_reservation_specification
• launch_template
• enclave_options
• host_id
• timeouts

Removed variables

• instance_count
• subnet_ids
• private_ips
• use_num_suffix
• num_suffix_format

Renamed outputs

• :info: All outputs used to be lists, and are now singular outputs due to the removal of count

Removed Outputs

• availability_zone
• placement_group
• key_name
• ipv6_addresses
• private_ip
• security_groups
• vpc_security_group_ids
• subnet_id
• credit_specification
• metadata_options
• root_block_device_volume_ids
• ebs_block_device_volume_ids
• volume_tags
• instance_count

Motivation and Context

• with the introduction of module level count and for_each, this means users can create as many ec2 instances as necessary with the module as opposed to the module itself trying to control how many instances are created. in addition, the use of for_each is preferred to avoid disruptive updates that can be encountered when using count and modifying one of the resources within the list

Closes #92
Closes #160
Closes #173
Closes #188
Closes #213
Closes #230

Breaking Changes

• Yes, see UPGRADE-3.0.md for upgrade instructions

How Has This Been Tested?

• I have tested and validated these changes using one or more of the provided examples/* projects
  • examples updated and validated for new changes

[antonbabenko]

Great to see this amount of improvements!

[bryantbiggs]

my local terraform-docs is @ v0.15 and hence the diff in the lint check 🤷🏼‍♂️

[antonbabenko]

my local terraform-docs is @ v0.15 and hence the diff in the lint check 🤷🏼‍♂️

I stopped managing terraform-docs with homebrew and downloaded 0.13 which we use everywhere :)

[antonbabenko]

LGTM

[antonbabenko]

Output for lazy people :)

[antonbabenko]

I fixed the docs, is there anything else in this PR?

[bryantbiggs]

Thank you! No, I think we should be all set - I'll keep an eye in case any issues start popping up

[antonbabenko]

Cool, v3.0.0 has been just released.

[firke]

hey @bryantbiggs @antonbabenko, thank you for your work on this! Quick question in regards to this release - what was the reasoning behind removing some of the outputs, e.g. availability_zone/private_ip?

[antonbabenko]

The main reason for that is that those attributes (outputs) are not in the resource itself (see aws_instance docs).

We probably had them in the past but they were actually arguments (inputs).

[firke]

@antonbabenko so I think Hashicorp have changed layout of docs a little, and outputs block does not explicitly mention all of the outputs, as it also includes inputs as part of outputs. This is listed in outputs part of docs:
In addition to all arguments above, the following attributes are exported
This would include private_ip as part of output.

I've tested it with a forked version - https://github.com/firke/terraform-aws-ec2-instance
Only private_ip output was added as part of test. Once used, private_ip attribute was available to other resources, route53 record as an example:

resource "aws_route53_record" "private_dns" {
  zone_id = data.terraform_remote_state.network.outputs.private_zone_id
  name    = "rpc"
  type    = "A"
  ttl     = "300"
  records = [module.ec2.private_ip]
}

[antonbabenko]

You are absolutely right, you could always use arguments as outputs. And we sometimes did this (and maybe still do) to implement poor-man depends_on solutions but this is not needed in this module.

[firke]

Ah, I see, this clicked with me just recently about the docs. So using same example of private_ip, referencing which should be a trivial task, my understanding is just now without these outputs it's not possible, and if these are not returned will either need to use forked version or use a data source for an ec2_instance object? Sorry if I've missed something. :)

Added: slight clarification as well - i'm not setting private_ip as part of module, but it is still available as output.

[antonbabenko]

You're right about using the data-source aws_instance as a fetcher for the data you need based on EC2 instance id.

[suppix]

@antonbabenko Looks like it's not so handy, especially to use this module with terragrunt.

[antonbabenko]

This module creates an EC2 instance so that it is possible to get some more information about the resource using data source in other places. In terragrunt you can use generate block to create extra data-source code or make a small module that will act as a fetcher/bridge.

[github-actions]

I'm going to lock this pull request 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 related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.