At the moment there is a known issue with CPU instances on Crusoe Cloud where networking may not be functioning at boot. This impacts SSH/Terraform and the bootstraps scripts. If the Terraform below fails contact [email protected] for help.
Modify the locals in the main.tf resource YAML file.
locals {
my_ssh_privkey_path="</path/to/priv.key"
my_ssh_pubkey="<pub_key>"
worker_instance_type = "h100-80gb-sxm-ib.8x"
worker_image = "ubuntu22.04-nvidia-sxm-docker:latest"
ib_partition_id = "6dcef748-dc30-49d8-9a0b-6ac87a27b4f8"
count_workers = 2
headnode_instance_type="c1a.8x"
deploy_location = "us-east1-a"
...
}
And then apply, to provision resources
terraform init
terraform plan
terraform apply