What this category’s data looks like
This category’s data primarily originates from orders and inventory in the enterprise ERP system, BOM lists, process routes, equipment capacity ledgers, staff shift schedules, work-in-progress ledgers, and supplier arrival records. Daily documents include sales orders, master production plans, material requirements plans, weekly and daily scheduling tables, production work orders, material shortage lists, and delivery change notices. Data update frequency follows daily, weekly, or real-time adjustments based on orders or abnormal events. Data fields include order numbers, material codes, process numbers, equipment capacity units, shift durations, material inventory quantities, and more. Some documents are linked to process parameters and technical standard requirements from process specifications.
Constraints during deployment and upgrade
Data sources for this category are scattered, spanning multiple systems such as ERP, equipment ledgers, and scheduling systems. During deployment, permissions and synchronization rules for multi-source data docking must be configured. Data update rhythm is flexible. After deployment, both real-time and scheduled synchronization modes must be supported. During upgrades, compatibility with legacy synchronization logic and adaptation to new data sources is required. Daily document formats vary, including structured tables and unstructured process descriptions. Corresponding parsing templates must be configured during deployment. Field units differ across sources, so preset field mapping rules must be established to prevent Q&A result deviations caused by unit mismatches.
How to set configurations
The following configuration items apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://fastgpt:yourpass@mongodb-host:27017/fastgpt?authSource=admin | This category requires storage of large volumes of scheduling work orders and process document data. Configuring an authenticated MongoDB connection ensures stability of data storage and queries |
REDIS_URL | redis://:redispass@redis-host:6379/1 | Used to cache frequently called process specification Q&A pairs and real-time scheduling data, reducing database query frequency to support real-time Q&A requirements |
DB_MAX_LINK | 15–25 | Concurrent requests for this category change dynamically with order insertions and abnormal events. Setting a reasonable connection count prevents database resource exhaustion |
STORAGE_S3_ENDPOINT | Fill in according to the access address of the enterprise object storage service provider | Used to store large files such as uploaded process specification documents and scheduling reports, supporting multi-format document parsing requirements |
FILE_TOKEN_KEY | Randomly generated 32-bit string | Used to encrypt file download links, preventing unauthorized access to sensitive documents containing process parameters and scheduling rules |
ROOT_KEY | Configure according to internal enterprise permission specifications | Used to manage the highest platform permissions, ensuring the security of access and configuration permissions for production-related data |
Three common misconfigurations
- Symptom: The citation display switch is disabled in Q&A configuration, but the result still forcibly displays citation content. Cause: The process specification document library for this category is not bound to the application, and the system enables full citation logic by default.
- Symptom: Tool calls only function in some scheduling scenarios, and cannot cover abnormal scenarios such as material shortages or order insertions. Cause: Trigger rules for tool calls are not bound to the business scenarios of this category, causing the model to fail to recognize query requirements for abnormal scheduling.
- Symptom: Uploaded process specification documents cannot be parsed or loaded normally. Cause:
STORAGE_S3_ENDPOINTandFILE_TOKEN_KEYare not configured correctly, resulting in insufficient file read permissions or incorrect address configuration.
How to confirm successful configuration
- Check the configuration of
MONGODB_URIandREDIS_URL, perform basic connectivity tests to confirm that the database and cache service can connect normally. - Upload a process specification document of this category, test whether the parsing function operates normally, and confirm that the file reading and storage link has no abnormalities.
- Configure a test query, such as "What are the process parameters corresponding to the current work order", verify whether the Q&A result is associated with the content of the bound document library.
- Simulate abnormal scenarios such as order insertions or material shortages, test whether tool calls trigger the corresponding process standard query logic.
The values provided here are common starting points and should be evaluated against the reader's own samples.