<\/span><\/h3>\nFor production use, I want to adjust some important configurations. First, I check the .env<\/code> file in the docker directory:<\/p>\nCommand:<\/strong> cat docker\/.env<\/code><\/p>\nHere you can configure important settings such as the HTTP port, MySQL passwords, and MinIO passwords. By default, RAGFlow runs on port 80. If you want to use a different port, you can edit the docker-compose.yml<\/code> file:<\/p>\nCommand:<\/strong> nano docker\/docker-compose.yml<\/code><\/p>\nLook for the line 80:80<\/code> and change it to <YOUR_PORT>:80<\/code>, e.g., 8080:80<\/code> for port 8080.<\/p>\nAnother important point: RAGFlow uses Elasticsearch as the document engine by default. If you want to use Infinity instead (a faster alternative), you must first stop all containers:<\/p>\n
Command:<\/strong> docker compose -f docker-compose.yml down -v<\/code><\/p>\nWarning:<\/b> The -v<\/code> parameter deletes the Docker container volumes, and all existing data will be lost. Make sure you really want to delete all data before running this command.<\/p>\nThen set it in the .env<\/code> file:<\/p>\nCommand:<\/strong> echo \"DOC_ENGINE=infinity\" | sudo tee -a docker\/.env<\/code><\/p>\nAnd restart the containers:<\/p>\n
Command:<\/strong> docker compose -f docker-compose.yml up -d<\/code><\/p>\nFor my two RTX A6000 GPUs, I already enabled GPU acceleration in Phase 3. If you haven’t done that yet, you can do it now:<\/p>\n
Command:<\/strong> echo \"DEVICE=gpu\" | sudo tee -a docker\/.env<\/code><\/p>\nAnd restart the containers:<\/p>\n
Command:<\/strong> docker compose -f docker-compose.yml restart<\/code><\/p>\n<\/span>Phase 6: Upload and Analyze First Documents<\/span><\/h3>\nAfter RAGFlow has been successfully started, I can now upload and analyze the first documents. I open RAGFlow in the browser and log in (you have to create an account on the first start).<\/p>\n
The RAGFlow user interface is very intuitive:<\/p>\n
\n- \n
Knowledge Bases:<\/strong> Here you create knowledge bases for your documents<\/p>\n<\/li>\n- \n
Upload Documents:<\/strong> Here you can upload PDFs, Word files, Excel files, images, and more<\/p>\n<\/li>\n- \n
Chat:<\/strong> Here you can ask questions about your documents<\/p>\n<\/li>\n- \n
Settings:<\/strong> Here you configure LLM settings, chunking templates, and more<\/p>\n<\/li>\n<\/ul>\nFor a first test, I create a new knowledge base and upload a PDF document. RAGFlow processes the document automatically, creates chunks, and indexes the content. After processing, I can ask questions about the document and receive answers with source citations.<\/p>\n
Tip:<\/b> RAGFlow supports different chunking templates that you can select depending on the document type. For technical documents, I recommend the “Technical Document” template; for general texts, the “General” template.<\/p>\n