Skip to content

fix(modeldeployment): use v1beta1 InferenceSet and allow empty instanceType - #244

Merged
rambohe-ch merged 1 commit into
mainfrom
fix/modeldeployment-v1beta1-byo
Oct 19, 2026
Merged

rambohe-ch merged 1 commit into
mainfrom
fix/modeldeployment-v1beta1-byo

Conversation

@zhuangqh

Copy link
Copy Markdown
Contributor

Reason for Change:

KAITO's v1alpha1 InferenceSet webhook rejects spec.template.resource.instanceType when the operator runs with nodeProvisioner=byo. kaito-project/kaito#2438 relaxed this only for v1beta1, so ModelDeployments on BYO clusters fail with:
instanceType must be empty when nodeProvisioner is byo

  • Render the InferenceSet as kaito.sh/v1beta1 (served since KAITO v0.11, storage version).
  • Allow an empty instanceType: drop minLength from the values schema and omit the field (resource: {}) when unset.
  • Update README/values docs and e2e InferenceSet GVR/GVKs to v1beta1.

Verified with helm template/lint and server-side dry-run against a BYO AKS cluster running KAITO 0.13.0-rc.2: v1beta1 with instanceType is admitted, while v1alpha1 reproduces the rejection.

Requirements

  • added unit tests and e2e tests (if applicable).

Issue Fixed:

Notes for Reviewers:

…ceType

KAITO's v1alpha1 InferenceSet webhook rejects spec.template.resource.instanceType
when the operator runs with nodeProvisioner=byo. kaito-project/kaito#2438 relaxed
this only for v1beta1, so ModelDeployments on BYO clusters fail with:
  instanceType must be empty when nodeProvisioner is byo

- Render the InferenceSet as kaito.sh/v1beta1 (served since KAITO v0.11, storage version).
- Allow an empty instanceType: drop minLength from the values schema and omit the
  field (resource: {}) when unset.
- Update README/values docs and e2e InferenceSet GVR/GVKs to v1beta1.

Verified with helm template/lint and server-side dry-run against a BYO AKS cluster
running KAITO 0.13.0-rc.2: v1beta1 with instanceType is admitted, while v1alpha1
reproduces the rejection.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Qinghui Zhuang <qzhuang@microsoft.com>
@rambohe-ch
rambohe-ch merged commit 4bcb1dc into main Oct 19, 2026
13 checks passed
@zhuangqh
zhuangqh deleted the fix/modeldeployment-v1beta1-byo branch October 19, 2026 09:47

This branch was successfully deployed

1 active deployment
e2e-test — 7505e55a Deployed Oct 19, 2026 by zhuangqh via e2e-tests #644
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.

2 participants