Skip to content

runtime: Conform to static_resource_mgmt behavior for templating and allow one-time resize at restore time. #464

Open
Camelron wants to merge 2 commits into
saul/msft-previewfrom
cameronbaird/saul-preview/factory-static-enlightenment
Open

runtime: Conform to static_resource_mgmt behavior for templating and allow one-time resize at restore time. #464
Camelron wants to merge 2 commits into
saul/msft-previewfrom
cameronbaird/saul-preview/factory-static-enlightenment

Conversation

@Camelron

@Camelron Camelron commented Jul 9, 2026

Copy link
Copy Markdown

This has two staged changes on top of the upstreamed clh-go templating support PR #461

  1. Hotplugging memory into VMs with zoned memory. This enables us to resize a templated VM to the workload size. Should also be cherry-picked and sent upstream.
  2. Msft-only change to hook up the fork's static_resource_* configs to the template paths. Upstream doesn't have/use these configs.

End result is that we can have default_* configs = 0, static_resource_management=true, AND VM templating enabled. This means we get VM templating with no changes to our VM size model.

In the templating configuration we can restore from the default size (512Mi) and resize UP to match the deployment's required size. What we do lose is the ability to deploy with limits less than the default size (< 512Mi).

Camelron added 2 commits July 9, 2026 20:08
For CLH, handle larger sizes for incoming deployments by resizing
the VM inherited from the template VM. i.e. incoming pod is
1024M and the template VM is 512M? Plug in a zone of size 512M.

Assisted-By: Claude Opus 4.8
Signed-Off-By: Cameron Baird <cameronbaird@microsoft.com>
This patch is for the msft fork only. It wires up the static size
configs such that factory respects the static_ fields.

This means we have the same VM size model as is in production today:
1. With no limits specified, the user gets the static size 512Mi,1vcpu
2. With limits specified, the user gets the exact specified size
	which is limits.memory,limits.vcpus for the deployment

Assisted-By: Clause Opus 4.8
Signed-Off-By: Cameron Baird <cameronbaird@microsoft.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant