This document is based on FastGPT v4.17.0.
What This Category's Data Looks Like
Daily documents for this category include process sheets, CAD patterns, layout plans, cutting orders, first sample seals, process quality sheets, and packaging size matching sheets. Data sources cover orders and size charts, CAD design files, fabric and accessory inspection records, hanging system process logs, needle detector test data, and customer sample sealing requirements. Data updates are performed per order, per style, and per batch. Single batch data includes fields such as cutting piece number, size, color batch, and process node. Cutting pieces are irreversible, manual processes are intensive, there are many styles, and seasonality is strong.
Constraints Imposed on Deployment and Upgrade by These Characteristics
The multi-source heterogeneous data and batch-based update characteristics of this category require deployment to support index isolation by business unit, to avoid mixing data from different orders. The irreversible nature of cutting pieces means traceability must be precise to individual cutting pieces or process nodes, so deployment must support fine-grained vector retrieval configuration. The strong seasonality and large number of styles require upgrades to allow quick switching of knowledge base groups, reducing downtime from full index reconstruction. At the same time, multi-device access in the workshop intranet environment requires deployment configuration to adapt to intranet domain names and port restrictions, ensuring normal system access at each workstation.
How to Configure Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://fastgpt:your-password@internal-mongo:27017/fastgpt?authSource=admin | Data for this category is stored per order batch. A fully authenticated connection string must be used to support isolation of multiple datasets. |
DB_MAX_LINK | 20–30 | Single batch data volume fluctuates greatly. Avoid connection pool exhaustion affecting concurrent traceability queries. |
REDIS_URL | redis://:your-redis-password@internal-redis:6379/0 | Real-time process data from hanging systems and needle detectors must be cached to ensure fast retrieval response. |
FE_DOMAIN | http://internal-fastgpt:3000 | Adapt to the workshop intranet access environment to avoid cross-domain access restrictions. |
STORAGE_S3_ENDPOINT | http://internal-storage:9000 | CAD patterns and layout plan files for this category have large file sizes. Intranet storage reduces upload and download latency. |
Three Common Misconfigurations
- Symptom: After starting Docker, the image status is up, but the frontend port 3000 cannot be accessed for login. Logs prompt MongoDB connection failure, and manual verification of the MongoDB account password fails. Cause: The
MONGODB_URIconfiguration is not set to a complete connection string with correct authentication information, or the intranet firewall blocks access to the MongoDB port. - Symptom: After configuring a model, clicking test prompts an error. Forcing the error to be ignored allows the system to run, but timeouts occur during batch traceability. Cause: The
DB_MAX_LINKconfiguration value is too low, leading to connection pool exhaustion during concurrent queries for multiple orders. - Symptom: After importing a process sheet, search results cannot be associated with the corresponding cutting piece number. Cause: An independent index is not created per business batch, or the vector retrieval granularity does not match the cutting piece number field.
How to Confirm the Configuration Is Correct
- Log in to the configured intranet frontend address, use the preset account to complete login, and verify that there are no permission or connection errors.
- Manually import a cutting order file, initiate a single cutting piece traceability query, and confirm that matching process quality sheet data is returned.
- Check the MongoDB connection log to confirm that the real-time connection count does not exceed the threshold set by the
DB_MAX_LINKconfiguration. - Upload a CAD pattern file, confirm that the file can be stored normally and associated with the corresponding order batch.
The values provided here are common starting points, and should be evaluated against deployment samples specific to the target environment.