offline/online serving is stuck at fetching files in paraworkers

[hyuenmin-choi]

Hello guys I have some problems with deploying distserve on docker.

This is command what i used to launch docker env.

docker run --gpus all -it --name dist_bench --network=host --shm-size=10g -v /home/hyunmin/dataset:/workspace/dataset nvcr.io/nvidia/pytorch:24.05-py3 bash

I succefully build DistServe & SwiftTransformer inside my docker too.

However, when I tried to launch distserve online api_server or offline llm inference.
It just stucks in initializing the engine.

Is this problem related to shm size options?

Can anyone share the env or commands which successfully launch the serving.

Thank you in advance.