Reference only; verify against the target FastGPT release and deployment documentation.
What the data for this category looks like
Data sources involved in document and material drafting for non-ferrous metal smelting include production shift reports, grade test reports issued by quality inspection centers, process procedure manuals, and procurement and sales contract documents. Data updates are completed per production shift or daily. Single document length varies widely, from shift logs of hundreds of characters to process optimization plans of thousands of characters. Documents include fixed fields such as furnace body number, metal grade, smelting duration, and unit energy consumption. Common field units are tons, %, hours, and kilowatt-hours.
What constraints these characteristics impose during deployment and upgrade
Scattered data sources covering production, quality inspection, business and other document types require configuration of cross-source data access storage connection parameters during deployment. The wide range of document lengths, from short logs to long plan documents, requires vector database parameters adapted to different text processing length requirements. Data updates per shift or daily require retaining incremental synchronization configuration entries during upgrade, to avoid excessive system resource usage from full re-import. Clear requirements for fixed fields and units require presetting field mapping rules during knowledge base construction, to ensure accurate field matching during subsequent retrieval.
Configuration settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://fastgpt:xxxx@internal-host:27017/nonferrous_metal?authSource=admin | Stores metadata and retrieval indexes for smelting documents, adapting to structured storage requirements for multiple document types |
REDIS_URL | redis://:redis-pass@internal-redis:6379/1 | Caches frequently accessed shift reports and quality inspection data, supporting shift-level data refresh |
DB_MAX_LINK | 20–30 | Matches peak load of concurrent retrieval of multiple documents in smelting scenarios, preventing retrieval failures due to connection pool exhaustion |
STORAGE_S3_ENDPOINT | Internal enterprise object storage service address | Stores large-volume process optimization plans and annual smelting reports, adapting to storage requirements for long documents |
ROOT_KEY | Randomly generated 32-character string | Configures the highest access permission for the knowledge base, ensuring access to smelting confidential documents is controllable |
Three common mistakes
- Phenomenon: After starting the Docker container, the access interface prompts model connection failure, and checking the logs shows curl connection timeout. Cause: Ollama and FastGPT containers were not started in order, and the Ollama service was not initialized before FastGPT started.
- Phenomenon: Empty results are returned when retrieving smelting documents, or field matching errors occur. Cause: Corresponding field mapping rules were not configured, causing the knowledge base to fail to recognize dedicated fields such as furnace number and grade.
- Phenomenon: The aiproxy container continuously reports errors and exits after docker-compose startup. Cause: The
STORAGE_S3_ENDPOINTparameter was not configured correctly, causing large-volume documents to fail to be stored normally, triggering storage verification failure.
How to confirm the configuration is complete
- Run the docker ps command to confirm that all FastGPT-related containers and dependent containers are in running state.
- Enter the FastGPT knowledge base management interface, upload a smelting shift report, and confirm that the document can be parsed normally and an index is generated.
- Configure a model test interface, enter a test question containing furnace number and grade fields, and confirm that matching document fragments can be returned.
- Check the background logs to confirm that there are no error messages for database connections or cache service connections.
The values given are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.