ChromaDB is a high-performance vector database designed to store and search large volumes of AI-processed text efficiently. While m-Power includes a built-in vector database suitable for testing and smaller documents, mrc strongly recommend installing ChromaDB when deploying your AI Assistants and Content Retrievers in a production environment. ChromaDB is optimized for larger files, multiple uploads, and faster search performance — making it the preferred choice for live environments where accuracy and reliability are essential.
Windows Installation
Follow these steps to install ChromaDB on Windows:
- Download and Extract:
- Download the ChromaDB package.
- Unzip ChromaDB.zip to a temp folder (e.g. C:\temp).
- Move the extracted ChromaDB folder to C:.
- Install Python:
- Download the Python installer here.
(Note: Python 3.13 is the supported version for ChromaDB.) - Move the .exe to C:\ChromaDB.
- Right-click the installer and choose “Run as Administrator”.
- Check “Add to PATH” and “Use admin privileges”.
- Choose “Customize Install”, keep defaults, and set install path to:
C:\ChromaDB\Python313
- Download the Python installer here.
- Install ChromaDB:
- Open Command Prompt as Administrator.
- Type:
pip install chromadb
- Verify Configuration:
- Edit C:\ChromaDB\Chroma-DB-Service.xml.
- Verify the Python path (line 6) matches the install location from step 2.
- Change the port (line 7) if port 8000 is already in use.
- Install Windows Service:
- Open Command Prompt as Administrator.
- Type:
cd C:\ChromaDB - Type:
Chroma-DB-Service.exe install
- Start and Verify Service:
- Start the ChromaDB service from Windows Services.
- If it stops or fails, check logs in C:\ChromaDB\logs.
- Open http://localhost:8000/docs to confirm ChromaDB is running.
Configure m-Power
Once ChromaDB is installed and running, configure m-Power to use it as the vector database:
- Open the following file:
/m-power/mrcjava/WEB-INF/classes/mrc-runtime.properties - Add the following lines:
embedding_store_type=chroma
chroma_url=http://localhost:8000 - Save the file and restart Tomcat.
Repeat the above steps in your production mrc-runtime.properties file as well.
Troubleshooting
If the ChromaDB service fails to start or stops unexpectedly:
- Check the logs at C:\ChromaDB\logs for detailed error messages.
- Ensure port 8000 is not already in use by another service.
- Confirm the Python path in C:\ChromaDB\Chroma-DB-Service.xml is correct.
- Verify ChromaDB was installed successfully by running:
pip show chromadb
If issues persist, reinstall Python and ChromaDB following the steps above.
Linux Installation
Follow these steps to install ChromaDB on Linux:
- Install Docker:
yum install docker - Install ChromaDB:
docker pull chromadb/chroma - When prompted, choose: docker.io/chromadb/chroma
- Start ChromaDB:
docker run -p 8000:8000 chromadb/chroma - Configure m-Power by following the steps above.