I self-hosted an AI support bot with FastGPT + Docker — 3 gotchas that will break your install

Description and use

The author deployed FastGPT on Linux with Docker, encountering three issues: MinIO image pull failure, incorrect CSV FAQ header, and missing embedding model. After fixes, the bot operates correctly.

Problem

Installing FastGPT v4.15 on a fresh Linux box with Docker fails due to missing MinIO image, incorrect CSV FAQ format, and lack of an embedding model.

Outcome

FastGPT deployed successfully after addressing three critical issues: using a China‑based MinIO mirror, correcting CSV FAQ header, and providing an embedding model.

Workflow

• Pull FastGPT Docker images
• Replace MinIO image with registry.cn-hangzhou.aliyuncs.com/fastgpt/minio
• Prepare CSV FAQ with header q,a,indexes
• Configure embedding model (e.g., DeepSeek)
• Start Docker containers
• Verify knowledge‑base creation

Evidence

The MinIO image no longer pulls from Docker Hub.

Source

CSV FAQ import needs a third column. The header must be q,a,indexes, not q,a, or the import is rejected with a format error.

Source

You need an embedding model, not just a chat model. Without one, you can't even create a knowledge base ("System not embedding model"). DeepSeek alone isn't enough.

Source