#!/usr/bin/env bash

set -euo pipefail
# 2. Estrai la chiave API
API_KEY=$(grep '^nvapi-' /home/azurian/.nvidia | tr -d '\r\n ')


CONTAINER_NAME="nim-qwen3-32b-spark"
IMAGE="nvcr.io/nim/qwen/qwen3-32b-dgx-spark:1.1.0-variant"
CACHE_DIR="/home/azurian/.cache/nim/qwen3-32b-dgx-spark"
PORT="8001"

mkdir -p "$CACHE_DIR"
chmod -R a+rwX "$CACHE_DIR"

docker rm -f "$CONTAINER_NAME" 2>/dev/null || true

echo "Avvio Qwen3-32B NIM per DGX Spark..."

docker run -d \
  --name "$CONTAINER_NAME" \
  --restart unless-stopped \
  --runtime=nvidia \
  --gpus all \
  --ipc=host \
  --shm-size=16G \
  --env-file "$ENV_FILE" \
  -e NIM_GPU_MEM_FRACTION=0.20 \
  -v "$CACHE_DIR:/opt/nim/.cache" \
  -p "${PORT}:8000" \
  "$IMAGE"

echo
echo "Container avviato: $CONTAINER_NAME"
echo "Log:"
echo "docker logs -f $CONTAINER_NAME"
echo
echo "Health:"
echo "curl http://localhost:${PORT}/v1/health/ready"
echo
echo "Models:"
echo "curl -s http://localhost:${PORT}/v1/models | jq"
echo