generated from kubernetes/kubernetes-template-project
-
Notifications
You must be signed in to change notification settings - Fork 180
Closed
Description
Must haves
- Refactor the vllm specific code to become model server agnostic #383
- The metrics refresh time might be much larger than the refreshMetricsInterval #99
- scheduling changes for lora affinity load balancing #423
- Support ext_proc FULL_DUPLEX_STREAMED mode #388
- Introduce ResolvedRefs Status Condition #190
- Replace "Ready" Condition with "Accepted" #444
- Make ModelName immutable #408
- Remove
EndpointPickerNotHealthy
condition from InferencePool Status #385 - Remove v1alpha1 API #401
- Creating multiple InferenceModels with the same ModelName causes inconsistencies #394
- Rename TargetPortNumber to PortNumber in ExtensionRef #376
- InferencePool should have nested status #379
- Contract between Envoy Extproc and EPP is incorrect. #298
Nice to have
- Code health and repo maintenance
- Address all vulnerabilities flagged on the published images #344
- Create a cmd top level directory and move main.go there #345
- Add license headers to all files #281
- Rearrange directory structure of the repo #348
- Create a separate pkg for the controllers #359
- Create a separate datastore pkg #358
- Clean up any more V(numeric) for log levels #307
- Testing
- Guides
- Features
- Metrics
- Optimizations
This is an initial list subject to changes after discussing with the community
Metadata
Metadata
Assignees
Labels
No labels