What this category of data looks like
Data for power generation equipment manufacturing declaration materials and file reviews primarily originates from equipment design drawings, manufacturing process specifications, qualification declaration documents, quality inspection reports, archived files, and similar sources. Updates are triggered by project initiation, manufacturing milestones, and qualification annual reviews, with no fixed cycle. Document structure includes structured metadata such as equipment model, manufacturing batch, compliance number, rated power, and unstructured long text such as process descriptions and review transcripts. Some files are CAD-format binary documents. Field units include kW, hours, pieces, and similar units.
What constraints do these characteristics impose on deployment and upgrade
Reference only; verify against the target FastGPT release and deployment documentation.
How to Set Configuration Parameters
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
STORAGE_S3_ENDPOINT | Endpoint address adapted for internal object storage, e.g. http://192.168.1.100:9000, set single file limit to 500 MB | Declaration materials for power generation equipment include large CAD drawings and process documents, with large individual file sizes |
DB_MAX_LINK | 20–30 connections | Declaration materials for this category involve multi-table linked structured data, requiring sufficient database connection pool support for concurrent review requests |
MONGODB_URI | Authenticated connection string, e.g. mongodb://user:pass@host:27017/fastgpt?authSource=admin | File review data requires unstructured storage archiving, and MongoDB is suitable for semi-structured file metadata |
REDIS_URL | Password-protected standalone Redis instance address, e.g. redis://:pass@host:6379/0 | Need to cache review status and batch task progress to reduce database load |
FILE_TOKEN_KEY | 32-character random string, calibrated via actual testing | Need to securely generate file access tokens to prevent unauthorized access to classified declaration materials |
FE_DOMAIN | Public or internal domain for deployment service, e.g. https://fastgpt.manufacture.gov.cn | Need a unified frontend access entry to comply with internal office network domain rules |
Three Common Configuration Mistakes
- Phenomenon: Running the
docker pull fastgptcommand returns a pull timeout or image not found error. Reason: Domestic mirror acceleration source is not configured, or an expired image tag is used. - Phenomenon: An internal server error prompt appears when importing review-specific plugins. Reason: The
DB_MAX_LINKconfiguration value is too low, and concurrent import requests exhaust database connection pool resources. - Phenomenon: Server disk usage continues to grow, and knowledge base disk usage details cannot be queried. Reason: Storage paths for original files, split chunks, and embedding vectors are not differentiated, and no regular redundant data cleanup rules are configured.
How to Confirm the Configuration is Correct
- Run the
docker pscommand to confirm that the FastGPT container and dependent MongoDB and Redis containers are all running. - Upload a large CAD drawing under 500 MB, and confirm that the file upload succeeds and the corresponding metadata can be retrieved in the knowledge base.
- Initiate a batch file review task, and confirm that task progress is displayed in real time in the backend with no database connection timeout errors.
- Access the domain configured in
FE_DOMAIN, and confirm that the frontend page loads normally with no resource load failure prompts.
The values provided 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.