Final Fix: Robust Docker Multi-Stage and Dynamic Port for Hugging Face

Dockerfile

@@ -1,62 +1,45 @@
-# Stage 1: Build the application
+# Stage 1: Build the full application
 FROM node:20-bookworm AS builder
 
+# Set build-time directory
 WORKDIR /app
 
-# Copy root configurations
-COPY package.json package-lock.json ./
-COPY packages/shared/package.json ./packages/shared/
-COPY packages/server/package.json ./packages/server/
-COPY packages/client/package.json ./packages/client/
-
-# Install all dependencies for building
-RUN npm ci
-
-# Copy source code
+# Copy all source files
 COPY . .
 
-# Build all packages (shared -> server -> client)
-RUN npm run build
+# Install all dependencies and build
+RUN npm ci && npm run build
 
 
-# Stage 2: Runtime environment
+# Stage 2: Final Runtime Image
 FROM node:20-bookworm-slim
 
-# Install runtime dependencies: Redis, Java (for execution), and Python
+# Install runtime dependencies: Redis, Java, and Python
 RUN apt-get update && apt-get install -y \
     python3 \
     openjdk-17-jdk \
     redis-server \
     && rm -rf /var/lib/apt/lists/*
 
+# Set runtime working directory
 WORKDIR /app
 
-# Copy production manifests
-COPY package.json package-lock.json ./
-COPY packages/shared/package.json ./packages/shared/
-COPY packages/server/package.json ./packages/server/
-COPY packages/client/package.json ./packages/client/
-
-# Install only production dependencies
-RUN npm ci --omit=dev
-
-# Copy compiled results from the builder
-COPY --from=builder /app/packages/shared/dist ./packages/shared/dist
-COPY --from=builder /app/packages/server/dist ./packages/server/dist
-COPY --from=builder /app/packages/client/dist ./packages/client/dist
+# Copy EVERYTHING from builder (including dist and node_modules) 
+# to ensure workspace symlinks and dependencies are preserved exactly.
+COPY --from=builder /app .
 
-# Copy startup script
-COPY scripts/start-hf.sh ./scripts/start-hf.sh
+# Ensure the startup script is executable
 RUN chmod +x ./scripts/start-hf.sh
 
 # Environment variables for Hugging Face
 ENV NODE_ENV=production
-ENV PORT=7860
+# Hugging Face usually provides PORT, but we default to 7860
+ENV PORT=7860 
 ENV CLIENT_URL=*
 ENV REDIS_URL=redis://localhost:6379
 
 # Expose the HF port
 EXPOSE 7860
 
-# Start Redis and the server
+# Use the robust startup script
 CMD ["./scripts/start-hf.sh"]

packages/server/src/config/env.ts

@@ -34,7 +34,7 @@ function required(key: string): string {
 
 export const config: EnvConfig = {
   NODE_ENV: (process.env.NODE_ENV || 'development').trim(),
-  PORT: process.env.NODE_ENV === 'production' ? 7860 : parseInt(process.env.PORT || '3001', 10),
+  PORT: parseInt(process.env.PORT || '7860', 10),
   CLIENT_URL: (process.env.NODE_ENV === 'production' ? '*' : (process.env.CLIENT_URL || 'http://localhost:5173')).trim(),
   GLM_API_KEY: required('GLM_API_KEY'),
   GLM_BASE_URL: (process.env.GLM_BASE_URL || 'https://open.bigmodel.cn/api/paas/v4/').trim(),

scripts/start-hf.sh

@@ -1,13 +1,25 @@
 #!/bin/bash
-# Start Redis in the background
-redis-server --daemonize yes
+# Start Redis in the background with essential flags for containerized environments
+echo "Starting Redis server..."
+redis-server --daemonize yes --protected-mode no
 
-# Wait for Redis to be ready
-until redis-cli ping > /dev/null 2>&1; do
-  echo "Waiting for Redis..."
+# Wait for Redis to be ready (up to 30 seconds)
+COUNTER=0
+until redis-cli ping > /dev/null 2>&1 || [ $COUNTER -eq 30 ]; do
+  echo "Waiting for Redis ($COUNTER)..."
   sleep 1
+  ((COUNTER++))
 done
-echo "Redis is ready!"
+
+if [ $COUNTER -eq 30 ]; then
+  echo "⚠️ Redis failed to start or is unreachable, but continuing for app boot..."
+else
+  echo "✅ Redis is ready!"
+fi
+
+# Use the PORT environment variable if provided by HF, otherwise default to 7860
+export PORT="${PORT:-7860}"
+echo "🚀 Starting GLMPilot on port $PORT..."
 
 # Start the Node.js application
 npm run start -w packages/server