[Feature]: extended documentation for options

[stevekm]

Feature Type

Documentation

Seqera API Endpoint

No response

Feature Description

thanks for all the work on this Terraform provider, I just wanted to throw in this low-priority note that some of the docs could benefit from extra descriptions. Especially since it appears that some options have different or reversed options from the previous usages of tw and seqerakit, and especially for TowerForge options since they are so numerous.

Looking at the docs here; https://registry.terraform.io/providers/seqeralabs/seqera/latest/docs/resources/compute_env#nestedatt--compute_env--config--aws_batch--forge

some things that make me stop and scratch my head for a minute;

• bid_percentage (Number) : how are we representing the values for this? Is 90% supposed to be 90 or 0.9? I am assuming the former but an example would be good
• dispose_on_deletion (Boolean) : could use a reminder on how this works, especially since it sounds like the same setting as tw's preserve-resources but in reverse? Or is it not the same?
• fusion_enabled (Boolean) : does this also require that wave_enabled be set to true in the higher scope? Are there other settings required to be turned on / off for Fusion to work? Also, it seems like there is another Fusion setting here fusion2_enabled how are these related?
• gpu_enabled (Boolean) : what exactly does this do? Will I still need to configure a GPU-compatible AMI with [image_id](https://registry.terraform.io/providers/seqeralabs/seqera/latest/docs/resources/compute_env#image_id-1) (String) if I use this? And wouldnt I still need to add GPU-enabled EC2's to the instance types list?
• nvnme_storage_enabled (Boolean) : is this one a typo?

Perhaps there are actually two pieces to this; general docs on the settings for things like Fusion and TowerForge and then separate docs in the TF Provider that describes which Fusion, Wave, TowerForge etc. settings each Terraform Provider Setting is mapping back to.

Use Case

.

Proposed Terraform Configuration

Alternatives Considered

No response

Priority

Low - Nice to have

Additional Context

No response

[Shahbaz-mahmood123]

Hi @stevekm, this is a great list, thanks. The documentation is a big WIP at the moment, we have some plans for possibly decomposing some of the more complicated resources into their own resource as well, which would also help.

[gavinelder]

Hi @stevekm These are being documented but at a highlevel.

dispose_on_deletion - I believe it's the same as preserve-resources in TWCLI and matches the web-ui
"Disable to preserve the batch compute resources created by Seqera independent of the lifecycle of this compute environment"

Personally I have not fully tested this behaviour but it is being documented and tested.

bid_percentage - This is a whole number between 1 & 100 we now have validation checks to ensure it's an int along with updated documentation

enable_fusion - This now is documented fully in the provider with validation logic added to ensure appropriate wave config is setup.

gpu_enabled - This is to use the GPU optimized AMI from amazon - https://docs.aws.amazon.com/AmazonECS/latest/developerguide/ecs-optimized_AMI.html again this is to be documented.

nvnme_storage_enabled - This is a typo in our API which we can't change because it's breaking for consumers of it , however we have now updated the terraform provider as it inferred it's resource name from that.

[gavinelder]

Hi @stevekm sorry the significant amount of messages however this is the documentation we will provide on the topic.

• dispose_on_deletion (Boolean) When set to true for AWS Batch forge environments, automatically deletes AWS resources
  created during compute environment setup when the Terraform resource is destroyed.

The following AWS resources will be deleted:

1. AWS Batch Compute Environments - The Batch compute environment itself
2. AWS Batch Job Queues - Associated job queues (head queue, compute queue, dragen queue)
3. EC2 Launch Templates - Launch templates for the compute instances
4. IAM Roles - Execution roles, head job roles, and other service roles
5. IAM Instance Profiles - Instance profiles attached to compute instances
6. FSx File Systems - FSx for Lustre file systems (if created during forge)
7. EFS File Systems - Elastic File Systems (if created during forge)

Important: Deleting a workspace with active compute environments will bypass this cleanup
and require manual removal of AWS resources. We recommend deleting compute environments
before deleting workspaces.