Triton inference server hang when starting with MPS

AI & Data Science Data Processing
1

#inception
Description
When MPS is enabled, TRTIS hang after start, time out for inference request, but success message for checking server ready, model ready.
Triton Information
2.10.0 built in tritonserver:22.07-py3
T4, NV driver 470.141.03
Expected behavior
suppose triton server process inference request successfully or starting server fail