Configure Legacy Marker Parsing for FastGPT

Overview of Legacy Marker Parsing

This workflow enables custom document parsing for FastGPT versions released prior to V4.9.0, using the standalone Marker parsing service. It diverts supported document processing to a dedicated Marker container, rather than using FastGPT’s built-in parsing pipeline.

Deploy Marker Container Service

First, retrieve the official Marker container image with the following command:

docker pull crpi-h3snc261q1dosroc.cn-hangzhou.personal.cr.aliyuncs.com/marker11/marker_images:v0.1

After the image is downloaded, launch the container with required GPU acceleration and configuration:

docker run --gpus all -itd -p 7231:7231 --name model_pdf_v1 -e PROCESSES_PER_GPU="2" crpi-h3snc261q1dosroc.cn-hangzhou.personal.cr.aliyuncs.com/marker11/marker_images:v0.1

This command enables GPU acceleration via the --gpus all flag, maps host port 7231 to the container’s internal parsing port, names the running container model_pdf_v1, and sets PROCESSES_PER_GPU to 2 to limit concurrent parsing processes per attached GPU.

Configure FastGPT Environment Variables

Two mandatory environment variables must be adjusted in your FastGPT deployment to integrate with the Marker service:

• CUSTOM_READ_FILE_URL: The full endpoint for the custom parsing service. Replace the host portion (e.g., xxxx.com) with your deployed Marker service address; the path /v1/parse/file must remain unmodified. A sample valid value is http://xxxx.com/v1/parse/file.
• CUSTOM_READ_FILE_EXTENSION: A comma-separated list of file extensions that will be routed to the Marker parsing service. Multiple supported types must be separated by commas.

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.

Safety guardrails

Use [REDACTED_CREDENTIAL] for credentials and private data. Confirm the documented environment and version before review.

Rollback guidance

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