What this category of data looks like
Data for this category comes from PLM, ERP, MES, supplier materials, validation tests, and customer-specific requirements. Update frequency is irregular, triggered by each process or engineering change. Daily documents include ECN/ECR forms, change assessment reports, drawings and BOM version records, process validation reports, production switching plans, inventory disposal documents, and customer approval documents. Data fields include unique change identifier, version number, effective node, associated equipment list, approval status, etc. There are no unified fixed units; field units adjust according to the corresponding document type.
What constraints do these characteristics impose during deployment and upgrade
Data for this category is stored across multiple internal network systems. During deployment, database and storage services accessible via the internal network must be configured, and public network third-party resources cannot be relied upon. Data update frequency is irregular, and single update scale fluctuates with change complexity. Elastically scalable storage space must be reserved to adapt to fluctuations in change scale. Various document types have significant structural differences. During the deployment phase, adaptation to multi-format document parsing and vector index construction is required. The associated equipment list must be aligned with the equipment library used for equipment inspection and maintenance Q&A, and cross-system data synchronization rules must be configured. Changes involve customer-sensitive documents, so permission control logic must be configured during deployment to prevent unauthorized access.
How to set configurations
Based on FastGPT v4.17.0, the following configuration items need to be adapted to the business characteristics of this category:
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
MONGODB_URI | Internal network accessible MongoDB instance connection address, including authentication information with read-write permissions | Change documents and equipment association data for this category are stored in MongoDB, which needs to adapt to synchronous write requirements of multiple internal network systems |
REDIS_URL | Internal network Redis cache instance address, set a cache expiration time of 1800 seconds | Used to cache parsed results of change documents and equipment association relationships; 1800 seconds adapts to the complete processing cycle of a single change |
STORAGE_S3_ENDPOINT | Endpoint address of the internal network object storage service, configure a single file upload limit of 1000 MB | This category includes large documents such as drawings and validation reports, which need to adapt to the internal network storage environment; the 1000 MB limit covers the size of most change documents |
FE_DOMAIN | FastGPT front-end access domain name deployed on the internal network | Used to generate login-free links and load front-end resources, adapting to enterprise internal network access scenarios |
DB_MAX_LINK | First 20 associated documents | Limits the number of associated change documents returned per retrieval to avoid model processing failures caused by overly long context |
FILE_TOKEN_KEY | 32-bit random string used as the encryption key | Used to encrypt file access tokens, ensuring access security for sensitive content such as customer approval documents |
Three common configuration errors
- Phenomenon: After a login-free link user deletes a conversation, no residual records appear in background logs. Cause: Persistent storage rules for the Redis instance corresponding to
REDIS_URLare not enabled, causing cached session logs to be automatically cleared when the session is deleted. - Phenomenon: After deployment, more than 30 change management-related knowledge bases cannot be created. Cause: The default knowledge base quantity limit configuration of FastGPT is not adjusted, failing to adapt to the expansion requirements of change management knowledge bases for this category.
- Phenomenon: Drawings exported from PLM cannot be loaded after internal network deployment. Cause:
STORAGE_S3_ENDPOINTis not configured as the internal network storage address, and connection failure occurs when attempting to access a public network storage service.
How to confirm configurations are correctly applied
- Access the configured
FE_DOMAINwithin the internal network environment, verify that the FastGPT front-end page loads normally, and no error prompts related to public network access are displayed. - Upload an ECN document, verify that it is successfully stored in the storage service corresponding to the configured
STORAGE_S3_ENDPOINT, and no upload failure prompts are displayed. - Initiate a change Q&A related to equipment inspection, verify that returned results include associated BOM versions, process validation reports, and other documents, and the number of results matches the configured value of
DB_MAX_LINK. - Generate a login-free link and delete the corresponding conversation, verify that background logs still retain records of this operation, with no automatic clearing occurring.
The values provided are common starting points and should be evaluated against the samples of the respective deployment environments.