Add ollama-init service to auto-pull models on startup

Author: medevs

docker-compose.homelab.yml

@@ -1,23 +1,21 @@
 # =============================================================================
-# Docker Compose - Homelab Override (OPTIONAL)
+# Docker Compose - Homelab Override
 # =============================================================================
 # This file is for homelab/CI-CD deployments using GHCR images
 # Regular users should use docker-compose.yml directly
 # =============================================================================
 # Use with: docker compose -f docker-compose.yml -f docker-compose.homelab.yml up -d
 # =============================================================================
 
-version: '3.8'
-
 services:
   # ===========================================================================
   # Backend Service - Use GHCR Image
   # ===========================================================================
   backend:
-    # Remove build section and use GHCR image instead
+    # Override build with GHCR image
     image: ghcr.io/medevs/portfolio-backend:latest
-    # Keep all other settings from base compose
-    pull_policy: always  # Always pull latest image
+    build: !reset null
+    pull_policy: always
 
   # ===========================================================================
   # Frontend Service - Use GHCR Image
@@ -30,10 +28,17 @@ services:
   # ===========================================================================
   frontend:
     image: ghcr.io/medevs/portfolio-frontend:latest
+    build: !reset null
     ports:
       - "3100:3000"
     environment:
       # Only non-NEXT_PUBLIC_* runtime vars work here
       - NEXT_PUBLIC_ADMIN_API_KEY=${ADMIN_API_KEY}
     pull_policy: always
 
+  # ===========================================================================
+  # Ollama Init - Inherits from base compose
+  # ===========================================================================
+  # The ollama-init service from docker-compose.yml handles model pulling.
+  # No override needed - it will pull models defined in environment vars.
+  # ===========================================================================

docker-compose.yml

@@ -21,35 +21,35 @@ services:
       - APP_NAME=${APP_NAME:-AI Portfolio Backend}
       - APP_VERSION=${APP_VERSION:-1.0.0}
       - DEBUG=${DEBUG:-false}
-      
+
       # Server Settings
       - HOST=0.0.0.0
       - PORT=8000
-      
+
       # CORS Settings (comma-separated)
       - CORS_ORIGINS=${CORS_ORIGINS:-http://localhost:3000,http://localhost:3001}
-      
+
       # Ollama LLM Settings
       - OLLAMA_BASE_URL=${OLLAMA_BASE_URL:-http://ollama:11434}
       - OLLAMA_MODEL=${OLLAMA_MODEL:-llama3.2:3b}
-      
+
       # ChromaDB Settings
       - CHROMA_PERSIST_DIR=/app/data/chroma_db
       - CHROMA_COLLECTION_NAME=${CHROMA_COLLECTION_NAME:-portfolio_docs}
-      
+
       # Document Processing
       - UPLOAD_DIR=/app/data/documents
       - MAX_FILE_SIZE_MB=${MAX_FILE_SIZE_MB:-10}
       - ALLOWED_EXTENSIONS=${ALLOWED_EXTENSIONS:-.pdf,.md,.txt,.docx}
-      
+
       # RAG Settings
       - CHUNK_SIZE=${CHUNK_SIZE:-500}
       - CHUNK_OVERLAP=${CHUNK_OVERLAP:-50}
       - TOP_K_RESULTS=${TOP_K_RESULTS:-3}
-      
-      # Embedding Model
-      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-all-MiniLM-L6-v2}
-      
+
+      # Embedding Model (Ollama embedding model)
+      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-nomic-embed-text}
+
       # Security (REQUIRED - set in .env file)
       - ADMIN_API_KEY=${ADMIN_API_KEY}
     volumes:
@@ -62,13 +62,14 @@ services:
     networks:
       - portfolio-network
     depends_on:
-      - ollama
+      ollama-init:
+        condition: service_completed_successfully
     healthcheck:
       test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
       interval: 30s
       timeout: 10s
       retries: 3
-      start_period: 40s
+      start_period: 10s
 
   # ===========================================================================
   # Frontend Service (Next.js)
@@ -113,11 +114,11 @@ services:
     networks:
       - portfolio-network
     healthcheck:
-      test: ["CMD", "curl", "-f", "http://localhost:11434/api/tags"]
+      test: ["CMD-SHELL", "ollama list || exit 1"]
       interval: 30s
       timeout: 10s
       retries: 3
-      start_period: 60s
+      start_period: 30s
     # Optional: Set resource limits
     deploy:
       resources:
@@ -126,6 +127,62 @@ services:
         reservations:
           memory: 4G
 
+  # ===========================================================================
+  # Ollama Model Initialization (Pulls required models)
+  # ===========================================================================
+  # This init container ensures all required Ollama models are pulled before
+  # the backend starts. It runs once and exits successfully.
+  # ===========================================================================
+  ollama-init:
+    image: curlimages/curl:latest
+    container_name: portfolio-ollama-init
+    depends_on:
+      ollama:
+        condition: service_healthy
+    networks:
+      - portfolio-network
+    environment:
+      - OLLAMA_HOST=ollama:11434
+      - LLM_MODEL=${OLLAMA_MODEL:-llama3.2:3b}
+      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-nomic-embed-text}
+    entrypoint: ["/bin/sh", "-c"]
+    command:
+      - |
+        echo "=== Ollama Model Initialization ==="
+        echo "Waiting for Ollama to be ready..."
+
+        # Wait for Ollama API
+        until curl -sf http://$$OLLAMA_HOST/api/tags > /dev/null 2>&1; do
+          echo "Waiting for Ollama..."
+          sleep 2
+        done
+        echo "Ollama is ready!"
+
+        # Function to pull model if not exists
+        pull_model() {
+          MODEL=$$1
+          echo "Checking model: $$MODEL"
+
+          # Check if model exists
+          if curl -sf http://$$OLLAMA_HOST/api/tags | grep -q "\"name\":\"$$MODEL\""; then
+            echo "Model $$MODEL already exists, skipping pull"
+          else
+            echo "Pulling model: $$MODEL (this may take a while)..."
+            curl -X POST http://$$OLLAMA_HOST/api/pull -d "{\"name\": \"$$MODEL\"}" --no-buffer
+            echo ""
+            echo "Model $$MODEL pulled successfully!"
+          fi
+        }
+
+        # Pull LLM model
+        pull_model "$$LLM_MODEL"
+
+        # Pull embedding model
+        pull_model "$$EMBEDDING_MODEL"
+
+        echo "=== All models ready! ==="
+    restart: "no"
+
 # =============================================================================
 # Volumes (Data Persistence)
 # =============================================================================
@@ -145,4 +202,3 @@ volumes:
 networks:
   portfolio-network:
     driver: bridge
-