Files
transcription-api/Dockerfile.pytorch
2025-09-11 09:59:16 +02:00

69 lines
1.8 KiB
Docker

# Alternative: Use official PyTorch image with CUDA support
# This image has PyTorch, CUDA, and cuDNN pre-installed
FROM pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime
# Set timezone to avoid interactive prompts
ENV DEBIAN_FRONTEND=noninteractive
ENV TZ=UTC
# Install system dependencies
RUN apt-get update && apt-get install -y \
ffmpeg \
libsndfile1 \
git \
tzdata \
&& rm -rf /var/lib/apt/lists/*
# Set Python environment
ENV PYTHONUNBUFFERED=1 \
PYTHONDONTWRITEBYTECODE=1 \
PIP_NO_CACHE_DIR=1 \
PIP_DISABLE_PIP_VERSION_CHECK=1
# Set working directory
WORKDIR /app
# Copy requirements and install dependencies (excluding torch)
COPY requirements.txt .
RUN grep -v "^torch" requirements.txt > requirements_no_torch.txt && \
pip install -r requirements_no_torch.txt && \
pip install grpcio-tools==1.60.0
# Copy proto files and generate gRPC code
COPY proto/ ./proto/
RUN mkdir -p ./src && \
python -m grpc_tools.protoc \
-I./proto \
--python_out=./src \
--grpc_python_out=./src \
./proto/transcription.proto
# Note: Don't modify imports - they work as-is when sys.path is set correctly
# Copy application code
COPY src/ ./src/
COPY entrypoint.sh ./
RUN chmod +x ./entrypoint.sh
# Environment variables
ENV MODEL_PATH=large-v3 \
GRPC_PORT=50051 \
WEBSOCKET_PORT=8765 \
ENABLE_WEBSOCKET=true \
CACHE_DIR=/app/models \
TORCH_HOME=/app/models \
HF_HOME=/app/models
# Create model cache directory
RUN mkdir -p /app/models
# Volume for model cache
VOLUME ["/app/models"]
# Expose ports
EXPOSE 50051 8765
# Health check
HEALTHCHECK --interval=30s --timeout=10s --start-period=60s --retries=3 \
CMD python -c "import grpc; channel = grpc.insecure_channel('localhost:50051'); channel.channel_ready()" || exit 1
ENTRYPOINT ["/app/entrypoint.sh"]