33 lines
704 B
Bash
33 lines
704 B
Bash
#!/bin/bash
|
|
set -e
|
|
|
|
# Default environment variables
|
|
export MODEL_PATH=${MODEL_PATH:-"/repository"}
|
|
|
|
echo "Starting NeMo Skills inference endpoint..."
|
|
echo "Model path: $MODEL_PATH"
|
|
|
|
# Function to handle cleanup on exit
|
|
cleanup() {
|
|
echo "Cleaning up processes..."
|
|
kill $(jobs -p) 2>/dev/null || true
|
|
wait
|
|
}
|
|
trap cleanup EXIT
|
|
|
|
# Start the model server in the background
|
|
echo "Starting model server..."
|
|
ns start_server \
|
|
--model="$MODEL_PATH" \
|
|
--server_gpus=2 \
|
|
--server_type=vllm \
|
|
--with_sandbox &
|
|
|
|
# Start the HTTP endpoint
|
|
echo "Starting HTTP endpoint on port 80..."
|
|
python /usr/local/endpoint/server.py &
|
|
|
|
# Wait for both processes
|
|
echo "Both servers started. Waiting..."
|
|
wait
|