Deploy ChatGLM2 for FastGPT via Docker

Image and Port Specifications

Two official container images are provided for ChatGLM2 deployment. The global public image is stawky/chatglm2:latest. Users in mainland China can use the accelerated China mirror image: registry.cn-hangzhou.aliyuncs.com/fastgpt_docker/chatglm2:latest. The container listens on internal port 6006 by default; this port must be mapped to a host port to enable external connectivity for FastGPT integration.

Security Token Configuration

A default security token is pre-configured for the container: [REDACTED_CREDENTIAL]. This token functions as the channel key when integrating the ChatGLM2 model with OneAPI. If a custom security token is required, it can be set using the sk-key environment variable during container deployment, following standard Docker environment variable passing procedures.

Step-by-Step Deployment Workflow

1. Pull the selected container image. For global deployments: ``bash docker pull stawky/chatglm2:latest ` For mainland China users: `bash docker pull registry.cn-hangzhou.aliyuncs.com/fastgpt_docker/chatglm2:latest ``
2. Launch the container with the required port mapping. To use the default pre-configured security token, run: ``bash docker run -d -p 6006:6006 stawky/chatglm2:latest ``
3. To specify a custom security token instead of the default value, add the environment variable flag to the run command: ``bash docker run -d -p 6006:6006 -e sk-key=your-custom-security-token stawky/chatglm2:latest ``

Replace your-custom-security-token with your desired token value. The container will now run the ChatGLM2 model, accessible via the mapped host port 6006 for FastGPT integration.

Source: FastGPT official source

Applicability and version scope

Use this page for the documented Deployment and upgrades scenario. Confirm the FastGPT, dependency, API, and deployment versions in the official source before applying a change.

Rollback guidance

Restore the prior technical-content authority snapshot. Restore saved configuration and data snapshots, then repeat the smallest verification scenario.