When using multiple schedulingProfiles, EPP pods can start normally but fail to function properly

[learner0810]

What happened:

When using multiple schedulingProfiles, EPP pods can start normally but fail to function properly.

root@demo-master:~# curl  ${IP}:${PORT}/v1/completions     -H 'Content-Type: application/json'     -d '{
      "model": "Qwen/Qwen2-0.5B-Instruct",
      "prompt": "Write as if you were a critic: San Francisco",
      "max_tokens": 100,
      "temperature": 0
    }'
inference gateway: InferencePoolResourceExhausted - failed to find target pod: single profile handler is intended to be used with a single profile, failed to process multiple profiles

What you expected to happen:

The EPP pod failed to start and reported an error, indicating a configuration error in the schedulingProfiles.

How to reproduce it (as minimally and precisely as possible):

apiVersion: inference.networking.x-k8s.io/v1alpha1
kind: EndpointPickerConfig
plugins:
  - type: queue-scorer
  - type: kv-cache-utilization-scorer
  - type: prefix-cache-scorer
  - type: single-profile-handler
schedulingProfiles:
  - name: default
    plugins:
      - pluginRef: queue-scorer
      - pluginRef: kv-cache-utilization-scorer
      - pluginRef: prefix-cache-scorer
  - name: custom-profile
    plugins:
      - pluginRef: queue-scorer
      - pluginRef: kv-cache-utilization-scorer
      - pluginRef: prefix-cache-scorer
      - pluginRef: single-profile-handler

Anything else we need to know?:

Environment:

• Kubernetes version (use kubectl version):
• Inference extension version (use git describe --tags --dirty --always):
• Cloud provider or hardware configuration:
• Install tools:
• Others:

[learner0810]

/assign

[yankay]

Should it be /close

[nirrozenbaum]

Should it be /close

yes, thank you 🙏

closing as done, fixed by #1698