- Chat and configuration storage is managed using Docker volumes, ensuring data persistence regardless of the image version.
- There are various upgrade strategies that range from total manual control to complete automation using external tools.
- The architecture based on the combination of Open WebUI and Ollama allows absolute control over privacy and the power of local hardware.
If you've ever encountered a notification upon entering your AI interface that a new version is available, it's understandable to feel a bit apprehensive. Fortunately, there's a way to Update Open WebUI without losing anything (conversations, settings, etc.). Nobody likes losing their chat history or having to reconfigure everything from scratch, especially when you already have your templates and documents well organized on your Mac or PC.
The good news is that Open WebUI It's designed so that the software and data travel on separate paths. Basically, as long as you don't delete the Docker volume where the database is stored, you can change the application version as many times as you want without your chats disappearing into thin air.
Strategies for keeping Open WebUI up to date
Depending on how complicated you want things to be, there are several paths. If you're someone who prefers to have absolute control and doesn't want surprises, the ideal option is set a specific version (like version 0.8.6). This is highly recommended for production environments or work teams, as it allows you to read the release notes before making the switch.
On the other hand, if you only use it at home for experimenting, you can use the label :mainThis always points to the most recent version, which makes the process more agile, although you run the risk of missing some important changes. disrupt the functioning from some tool that you already had configured.
The manual update method
For those who prefer to do things at their own pace, the manual process is the most transparent way to update Open WebUI without losing anything. First, it is essential to perform a data backupbecause some database migrations are difficult to revert if something goes wrong. Once the files are secured, simply download the updated image using the command docker pull.
After downloading the new version, you must stop the current container and delete it. Don't panic. Your chats are not inside the containerbut on the associated volume. Finally, you run the command again. docker run original. An important trick: after updating, give it a clear browser cache Use Cmd+Shift+R to prevent the interface from looking strange due to old files.
Automating the process with Watchtower
If you find it a hassle to type commands every week, there's a tool called WatchtowerIt's a container that monitors for new versions of Open WebUI and, if so, automatically downloads and installs them without you having to lift a finger.
To run it just once and update everything instantly, you can run a quick command in the terminal. The best part is that Watchtower respects existing volumesTherefore, the update is done invisibly and your data remains intact. It's the preferred option for those looking for a "install and forget" experience.
How Open WebUI and Ollama work
Is it really possible to update Open WebUI without losing anything? You have to understand the architecture. Open WebUI is simply the visible face, the frontend that allows us to chat comfortably. The engine that actually powers the models is... Ollama, which handles local inference and manages GGUF files.
On Macs with Apple Silicon, the smartest thing to do is run Ollama natively to take advantage of GPU Metalwhile Open WebUI continues to run in Docker. These two communicate over Docker's internal network using the address host.docker.internal:11434achieving brutal performance without technical complications.
Powering AI with RAG and Functions
One of the reasons why it's worth keeping the system updated is to access features such as native RAGThis allows you to upload your own PDFs or text documents for the AI to use as a knowledge base. The system fragments the information and stores it in a vector database called ChromaDB.
Furthermore, updating Open WebUI without losing anything usually brings improvements to the Pipelines and FunctionsThese tools allow you to add Python logic directly to the chat, such as filters to remove email addresses from prompts or actions to automatically summarize text. Recently, the following has been integrated: Model Context Protocol (MCP), which allows AI to interact with external tools such as GitHub or Slack.
Docker alternatives and troubleshooting
Although Docker is the quick way, some users opt for a native installation using pip or by cloning the GitHub repository. This can improve resource efficiency by removing the virtualization layer, although it requires manually managing Python versions (3.11 being the most stable for this project).
If, after updating, you notice that the interface cannot find the Ollama models, it's most likely a network problem. On Linux, the flag is often missing. --add-host, while in other cases it may be that Ollama is linked only to 127.0.0.1preventing the container from accessing it. Configure the variable OLLAMA_HOST=0.0.0.0:11434 It usually solves 90% of these problems.
Updating Open WebUI without losing anything is, as you can see, very simple. Keeping the system up to date means taking advantage of the power of local models without sacrificing the convenience of a modern interface, always ensuring that data persistence is based on external volumes and not on the ephemeral container.
Editor specialized in technology and internet issues with more than ten years of experience in different digital media. I have worked as an editor and content creator for e-commerce, communication, online marketing and advertising companies. I have also written on economics, finance and other sectors websites. My work is also my passion. Now, through my articles in Tecnobits, I try to explore all the news and new opportunities that the world of technology offers us every day to improve our lives.