AKS Improvements #36
AKS Improvements #36
Conversation
github-actions
bot
added
documentation
| # - Pin operator/system components to the smaller cpu8 pool. | ||
| # - Expose two CPU-focused instance types aligned with the cpu8 and cpu16 node pools. | ||
|
|
||
| enableAnyscaleRayHeadNodePDB: true |
There was a problem hiding this comment.
Can you remove this yaml file?
We should just refer to the Anyscale Operator helm chart link and ask users to customize the values file themselves. We do not want to fork / maintain another helm values file per example.
| #checkov:skip=CKV_AZURE_227: "Ensure that the AKS cluster encrypt temp disks, caches, and data flows between Compute and Storage resources" | ||
|
|
||
| name = var.aks_cluster_name | ||
| name = local.cluster_name |
There was a problem hiding this comment.
Can you keep the original name unchanged?
| auto_scaling_enabled = true | ||
| min_count = 1 | ||
| max_count = 3 | ||
| min_count = 3 |
There was a problem hiding this comment.
Why change the min_count to 3? We should keep it as minimal as possible.
| min_count = 1 | ||
| max_count = 3 | ||
| min_count = 3 | ||
| max_count = 5 |
| min_count = 3 | ||
| max_count = 5 | ||
|
|
||
| upgrade_settings { |
There was a problem hiding this comment.
Can you remove this upgrade_settings change?
We prefer not to maintain any optional settings here. Users can customize on their own.
There was a problem hiding this comment.
with this one, if we don't provide it, I think the AKS rejects it - as in it has to have upgrade_settings..
| vm_size = "Standard_NC16as_T4_v3" | ||
| product_name = "NVIDIA-T4" | ||
| gpu_count = "1" | ||
| user_node_pools = { |
There was a problem hiding this comment.
Where are those GPU node pools? They are necessary.
There was a problem hiding this comment.
I will add them back on once I can test them, currently don't have GPU quota so couldn't test (WIP)
| @@ -1,13 +1,25 @@ | |||
| resource "random_string" "storage_suffix" { | |||
There was a problem hiding this comment.
Can you revert this change? We prefer not to have any randomness here. Otherwise debugging issues would be harder.
| default = "anyscale-operator" | ||
| } | ||
|
|
||
| variable "node_group_gpu_types" { |
There was a problem hiding this comment.
These are necessary and available across all examples.
| source = "hashicorp/azurerm" | ||
| version = "4.26.0" | ||
| } | ||
| random = { |
There was a problem hiding this comment.
Please revert this randomness change.
Pull request checklist
Please check if your PR fulfills the following requirements:
Pull Request Type
Does this introduce a breaking change?
Other information