What the data for this category looks like
Business data for this category comes primarily from order design systems, sheet and wood batch ledgers, CNC cutting and drilling device logs, ERP/MES systems, and moisture content testing equipment. Daily retained documents include order breakdown drawings, cutting lists, edge banding and drilling programs, coating process parameters, hardware accessory lists, and trial assembly and packaging records. Data updates trigger with orders, daily cycles, or per material batch. A single order maps to multiple sets of panel component data. Fields include panel dimensions, material batches, process parameters, and installation point information. Units include millimeters, cubic meters, kilograms, and others. Single batch data volume increases with the complexity of custom orders.
What constraints these characteristics impose on deployment and upgrade
This category has high data update frequency, and single orders link to multiple sets of panel component data. Deployment must adapt to high-frequency write and batch query loads, and reserve sufficient storage and computing resources. Multiple types of documents are accessed in mixed form, with large format differences between order breakdown drawings, process programs, and other files. Deployment must configure parsing rules adapted to different document structures to avoid parsing failures. Data links to multiple devices and system interfaces. Deployment must connect to docking links for ERP and CNC devices to ensure stable data synchronization. Upgrades must be compatible with historical batch sheet and process data. This avoids data format mismatches after upgrade, which would reduce the accuracy of equipment inspection and maintenance Q&A.
How to set configurations
The following configurations are based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 120–150 connections | This category has large single-batch panel component data volume. A stable database connection pool is required to support high-frequency queries and batch data imports |
REDIS_URL | Cache validity period of 3600 seconds | Sheet and process data update once daily or per batch. Match the cache validity period to the update rhythm to avoid query errors caused by expired data |
STORAGE_S3_ENDPOINT | Internal enterprise object storage access address | This category stores large volumes of documents such as order breakdown drawings and process programs. Connect to existing enterprise storage resources to ensure data security and access efficiency |
FILE_TOKEN_KEY | Set based on actual measurements | Different enterprises have different document access permission rules. Adjust the key configuration based on internal security policies |
ROOT_KEY | Replace with a randomly generated key in production environments | Avoid permission leakage risks caused by default keys, and comply with internal enterprise security management requirements |
Three common mistakes
- Phenomenon: After importing datasets such as order breakdown drawings and cutting lists, the system remains in the indexing state for a long time. Cause: Parsing rules for multiple types of documents in this category are not adapted. Parsing time for complex format process programs and order breakdown drawings exceeds the system default threshold.
- Phenomenon: The locally deployed API cannot be called by other internal applications. Cause: Cross-origin access rules are not configured correctly. External requests are intercepted by the system, and interface data for equipment inspection and maintenance Q&A cannot be obtained normally.
- Phenomenon: Authentication failure is returned when calling the local rerank model. Cause: The security credential of
requestAuthis not configured as a key matching the local model. The Bearer prefix is not automatically added, resulting in authentication failure.
How to confirm the configuration is correct
- Execute a test task for batch importing order breakdown drawings and cutting lists. Check the parsing progress and results to confirm that the data import process works normally.
- After configuring cross-origin access rules, use an internal application to send an API call request. Confirm that the interface can normally return Q&A results related to equipment inspection.
- Manually trigger a sheet batch data update. Confirm that the cache configuration takes effect, and data synchronization has no abnormalities.
- Verify the calling link of the local rerank model. Confirm that the authentication credential configuration is correct, and the returned sorting results meet business requirements.
The values listed here are common starting points, and should be evaluated against local deployment samples.