Hello all, Was wondering what these variables in this example.tfvars correspond to? They don't appear to be passed into the Metaflow module.

cpu_max_compute_vcpus     = 128
cpu_min_compute_vcpus     = 0
cpu_desired_compute_vcpus = 0

large_cpu_max_compute_vcpus     = 128
large_cpu_min_compute_vcpus     = 0
large_cpu_desired_compute_vcpus = 0

gpu_max_compute_vcpus     = 128
gpu_min_compute_vcpus     = 0
gpu_desired_compute_vcpus = 0

good catch, i think it may be a leftover from some early version where those vars had different names

Ok got it. Thanks very much @average-beach-28850! I think I'm on the right track then. I was getting worried that I missed some sort of configuration. 🙂

I'd also recommend using examples from https://github.com/outerbounds/terraform-aws-metaflow as a starting point these days, as they are more recent and more concise, not that tf config in metaflow-tools

Gotcha, thanks. Will do that. Is the sagemaker setup still valid in metaflow-tools? Need to setup sagemaker with access to Metaflow and that would be useful

it should be, but I'd still probably use it more as a reference. I don't beleive there is anything particularly creative with sagemaker there, aside from providing metaflow config vars

Hi @average-beach-28850, QQ - are these IAM roles/policies needed - https://github.com/outerbounds/metaflow-tools/blob/master/aws/terraform/metaflow/iam-custom-role.tf? What are these for? They're created outside of the metaflow terraform module. Thank you in advance!

its the "user" role that has sufficient permissions to submit jobs to Batch, read/write S3 etc. Its not used by anything "inside" the deployed stack but you may want to use it as a metaflow user (or create your own with a superset of those permissions)

Ahh ok got it, thank you.