vllm-project
Introduce wait time or retry before moving adapter to another pod if that pod is not ready
🚀 Feature Description and Motivation
This is a follow up issue from https://github.com/aibrix/aibrix/pull/257#issuecomment-2383841118
from @varungup90 "Here retry will be for model adapter reconcile. For 5 retries, with 5 second retry interval, pod on which adapter is scheduled is still not ready so lets reschedule the model adapter. Or another option is that track the time when pod is not ready for already scheduled adapter and from then even after 60 seconds pod is not ready then reschedule model adapter"
"goal is to not reschedule model adapter on first encounter of pod not ready"
Use Case
No response
Proposed Solution
No response
this won't affect functionality and would be a low priority item at this moment. I moved to v0.3.0
Is this issue still active? I just want to participate in this project. If possible, can you tell me about some of the code? This way I can integrate into this issue faster. @Jeffwan 😄
