Kubernetes and vllm Working Group
Details
We have had consistent zoom bombing problems. Link up 10m before start time
Many jobs require an opensource contribution to get in the door. LLMs can do github tickets.
vllm ticket review: https://github.com/vllm-project/vllm/issues/50128.
This ticket starts with comparing warm start vs. cold start. Someone found a path where there was a bug causing longer restarts because of tracing. The bug is called during the processing of the transformer backend where an AST is built an code is replaced with CUDA kernels. When a module is not found it generates another trace call which pushes on the stack and causes another layer of delays until it is garbage collected. The speedgraph shows this. Make sure to select the longest thread. Tricky
Speedgraph
###