#!/bin/bash conda run -n xinference --no-capture-output xinference-local > >(tee xinference-output.log) 2>&1 & PID=$! echo "Started xinference-local with PID $PID" echo "Checking if output.log has content..." while [ ! -s xinference-output.log ]; do echo "Waiting for output to appear in xinference-output.log..." sleep 1 done while ! grep -q "Uvicorn running on http://127.0.0.1:9997" xinference-output.log; do sleep 1 done echo "xinference-local started successfully" bash /root/start_models.sh bash /root/start_models_emb.sh echo "Started xinference-local with PID $PID" echo $PID > xinference.pid echo "model started successfully"