Initial commit - production deployment

services/training/Dockerfile

@@ -0,0 +1,70 @@
+# =============================================================================
+# Training Service Dockerfile - Environment-Configurable Base Images
+# =============================================================================
+# Build arguments for registry configuration:
+#   - BASE_REGISTRY: Registry URL (default: docker.io for Docker Hub)
+#   - PYTHON_IMAGE: Python image name and tag (default: python:3.11-slim)
+# =============================================================================
+
+ARG BASE_REGISTRY=docker.io
+ARG PYTHON_IMAGE=python:3.11-slim
+
+FROM ${BASE_REGISTRY}/${PYTHON_IMAGE} AS shared
+WORKDIR /shared
+COPY shared/ /shared/
+
+ARG BASE_REGISTRY=docker.io
+ARG PYTHON_IMAGE=python:3.11-slim
+FROM ${BASE_REGISTRY}/${PYTHON_IMAGE}
+
+WORKDIR /app
+
+# Install system dependencies including cmdstan requirements
+RUN apt-get update && apt-get install -y \
+    gcc \
+    g++ \
+    make \
+    curl \
+    build-essential \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy requirements
+COPY shared/requirements-tracing.txt /tmp/
+
+COPY services/training/requirements.txt .
+
+# Install Python dependencies
+RUN pip install --no-cache-dir -r /tmp/requirements-tracing.txt
+
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy shared libraries from the shared stage
+COPY --from=shared /shared /app/shared
+
+# Copy application code
+COPY services/training/ .
+
+
+
+# Add shared libraries to Python path
+ENV PYTHONPATH="/app:/app/shared:${PYTHONPATH:-}"
+
+# Set TMPDIR for cmdstan (directory will be created at runtime)
+ENV TMPDIR=/tmp/cmdstan
+
+# Install cmdstan for Prophet (required for model optimization)
+# Suppress verbose output to reduce log noise
+RUN python -m pip install --no-cache-dir cmdstanpy && \
+    python -m cmdstanpy.install_cmdstan
+
+# Expose port
+EXPOSE 8000
+
+# Health check
+HEALTHCHECK --interval=30s --timeout=10s --start-period=5s --retries=3 \
+    CMD curl -f http://localhost:8000/health || exit 1
+
+# Run application with increased WebSocket ping timeout to handle long training operations
+# Default uvicorn ws-ping-timeout is 20s, increasing to 300s (5 minutes) to prevent
+# premature disconnections during CPU-intensive ML training (typically 2-3 minutes)
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000", "--ws-ping-timeout", "300"]