What this category of data looks like
Data for this category comes primarily from product digital models, GD&T specifications, process route documents, coordinate measuring machine data, on-site trial feedback, and metrology systems. Daily documents include fixture requirement sheets, positioning datum schemes, fixture drawings, gauge operation manuals, acceptance reports, calibration records, change histories, and similar items. Data updates follow project milestones, product changes, and scheduled calibrations. Document fields include GD&T tolerance values, coordinate measuring machine measurement results, calibration cycles, version numbers, and similar items. Units include millimeters, newtons, hours, and similar units.
What constraints these characteristics impose on deployment and upgrade
Data sources for this category cover engineering digital models, metrology data, and on-site feedback. Mixed-format input requires deployments to support multiple file types, including CAD drawings and CSV metrology reports. Data updates have no fixed schedule, and follow project launches, product changes, and calibrations. Upgrade processes must support incremental vector database migration to avoid the time cost of full reconstruction. Documents include specialized tolerance fields and structured records. Deployment configurations must support embedding matching for specialized terminology, and also support version-associated storage of change histories.
How to set configurations
The following configurations are based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 200–300 connections | Adapts to the storage needs of multiple document associations per project for this category, and prevents database connection overflow |
MONGODB_URI | Dedicated replica set connection string | Stores structured business data such as change histories and calibration records, and ensures read and write stability |
STORAGE_S3_ENDPOINT | Domestic compliant cloud storage access address | Stores large-volume drawing files such as fixture drawings and gauge operation manuals, and supports multi-format upload requirements |
REDIS_URL | 127.0.0.1:6379 (local instance) or dedicated cache instance address | Caches frequently accessed GD&T specifications and common fixture parameters to reduce retrieval latency |
FILE_TOKEN_KEY | Randomly generated 32-bit hexadecimal string | Verifies the legitimacy of fixture documents uploaded to the knowledge base, and prevents unauthorized file access to the system |
PG_URL | Configured with a dedicated connection string for project databases | Stores structured text data such as process routes and positioning datum schemes, and supports complex queries |
Three common mistakes
- Issue: After deploying FastGPT via Docker, uploading fixture drawing knowledge base files fails, and the interface returns an upload timeout prompt. Cause: The local storage directory is not correctly mounted, or the storage permissions corresponding to
STORAGE_S3_ENDPOINTare not configured, resulting in failure to write files to the specified path. - Issue: After upgrading to a new version, calibration records and change history data in the old version's vector database cannot be retrieved normally. Cause: The incremental vector database migration script was not executed, and the version association fields of the original documents were not retained during direct full reconstruction.
- Issue: After upgrading to the latest version, using voyage index to retrieve fixture documents returns a 400 status code with no response body. Cause: The adaptation parameters of the index configuration were not updated, or the index shard size does not match the new system.
How to confirm configurations are properly set
- Upload a standard fixture drawing document to the knowledge base, and confirm that the file can be stored normally and a vector index is generated.
- Run a database connection test, and confirm that the number of connections does not exceed the threshold set by the
DB_MAX_LINKconfiguration. - Check the cache logs, and confirm that frequently accessed GD&T specification data has been cached normally.
- Attempt to upload an unauthorized non-fixture document, and confirm that the system intercepts this type of file.
The values provided here are common starting points, and should be measured against the reader's own samples.