You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It may not be necessary to create a gpu module if there are no resources unique to this configuration. In that case, we can conditionally overlay all gpu options in the node_pool module.
The text was updated successfully, but these errors were encountered:
https://github.com/equinix/terraform-metal-multiarch-k8s/blob/main/kubernetes-node-pool.tf could be implemented as a for loop over a var.node_pools slice of structs.
The existing hard-coded green/blue node pools would be replaced with user-defined sets of node pools where all options can be specified.
Variables that look like
count_arm = 1; plan_arm = "foo"
would becomenode_pools = [{name: "arm", count=1, instance_type: "foo"}]
.In order to get this benefit, we may need to combine gpu_node_pool module options and customizations into the node_pool module.
node_pool/modules/gpu/main.tf ()
node_pool/variables.tf gpu_node_options = { install_foo: true, ...}
It may not be necessary to create a gpu module if there are no resources unique to this configuration. In that case, we can conditionally overlay all gpu options in the node_pool module.
The text was updated successfully, but these errors were encountered: