Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Create a kserve auxillary controller to manage extra resources for LLM serving #58

Closed
Tracked by #70
VedantMahabaleshwarkar opened this issue Jul 18, 2023 · 3 comments
Assignees
Labels
odh-release/1.8 Need to do for ODH Release v1.8.0 rhods-1.32

Comments

@VedantMahabaleshwarkar
Copy link
Contributor

VedantMahabaleshwarkar commented Jul 18, 2023

We need a auxiliary controller for the LLM Serving stack. The needs for this controller might evolve over time but currently are as follows :

  • Define how we filter for LLM Serving model namespaces
    • Maybe some NS annotation similar to modelmesh
  • Watch LLM Serving enabled namespaces
    • Create SMMR and PeerAuthentication resources for the namespace when a new NS is detected with LLM Serving enabled.
    • Clean up SMMR and PeerAuthentication resources if NS is deleted / LLM Serving label is removed
@heyselbi heyselbi added the odh-release/1.8 Need to do for ODH Release v1.8.0 label Jul 19, 2023
@danielezonca danielezonca changed the title Create a kserve auxillary controller to manage extra resources for watsonx Create a kserve auxillary controller to manage extra resources for LLM serving Jul 21, 2023
@VedantMahabaleshwarkar
Copy link
Contributor Author

The auxillary controller will be odh-model-controller and it will be separated from it's current deployment structure (which is an overlay to modelmesh).

@VedantMahabaleshwarkar
Copy link
Contributor Author

The initial cut for the controller was created here : https://github.com/VedantMahabaleshwarkar/odh-model-controller/tree/kserve-metrics

On 8/4 new requirements for monitoring were added. See Req 7 in the Requirements Doc

These new requirements mean that the overall monitoring solution changes from Option A to Option C as described in this doc

The change in logic for the controller corresponding the the change from Option A to Option C is currently in progress

@VedantMahabaleshwarkar
Copy link
Contributor Author

#65

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
odh-release/1.8 Need to do for ODH Release v1.8.0 rhods-1.32
Projects
Status: No status
Status: No status
Status: Done
Development

No branches or pull requests

2 participants