What data for this category looks like
Data is sourced from multiple business systems of coating, resin, additive, agrochemical intermediate, and specialty chemical enterprises. These systems include formula and process libraries, raw material batch records, reactor DCS data, weighing system logs, LIMS inspection reports, and warehouse inbound and outbound documents. Daily generated documents include batch production instructions, feeding review sheets, reaction process records, cleaning records, deviation and change orders, and more. Data updates are tied to each production batch or key process step. Due to multi-variety production with batch switching, it is necessary to distinguish formulas, feeding amounts, and cleaning confirmation records for different batches. Some documents involve confidential formula content.
What constraints these characteristics impose during deployment and upgrade
Multi-source heterogeneous data sources require flexible support for docking industrial system interfaces with different protocols during deployment, to avoid hard-coded adaptations. Batch and key step-based data updates require incremental synchronization logic triggered by production batches, to avoid excessive system resource usage from full synchronization. Batch isolation requirements from multi-variety switching require database configurations that support batch-level permission isolation and data partitioning. Documents involving confidential formulas require encrypted storage and access control rules configured during deployment. Format differences in non-standard documents such as cleaning records and deviation reports require retaining flexible document parsing adaptation capabilities during upgrades, to adapt to document structure changes across different batches.
How to set configurations
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
DB_MAX_LINK | 10–15 connections | Single batch production in this category triggers multi-source data pulling. Balance database connection count and server memory usage, calibrated via actual testing |
REDIS_URL | redis://:[password]@[host]:6379/0 | Used to cache batch data and temporary parsing results, reducing system overhead from repeated data pulling |
FILE_TOKEN_KEY | 32-character random string | Used to generate file access tokens, ensuring download permissions for confidential documents |
STORAGE_S3_ENDPOINT | http://[internal storage host]:9000 | Adapts to enterprise internal object storage architecture, stores production documents and batch records |
MONGODB_URI | mongodb://[username]:[password]@[host]:27017/fastgpt?authSource=admin | Supports batch-level distributed data storage and retrieval |
ROOT_KEY | Custom complex key string | Used for administrator permission verification, ensuring system access security after deployment |
Three common mistakes
- Symptom: High persistent disk read/write activity occurs after Docker deployment, with rapid disk usage growth. Cause: The
DB_MAX_LINKparameter is not set, the database connection pool expands without limit, and disk cache files are read and written frequently. - Symptom: Module missing errors occur when starting a local non-Docker deployment. Cause: Dependent components adapted for industrial system data docking are not installed, making it impossible to pull reactor DCS and LIMS data sources.
- Symptom: Generated ledger reports contain mixed data across batches. Cause: Retrieval filtering rules based on production batches are not configured, resulting in retrieval results that do not distinguish documents from different production batches.
How to confirm the configuration is properly set
- Database connection logs can be checked to confirm that the connection count stays within the configured
DB_MAX_LINKrange, with no abnormal growth. - A ledger generation task for a single production batch can be triggered to confirm that only data source documents for the corresponding batch are pulled, with no cross-batch data mixing.
- Access permissions for confidential documents can be verified to confirm that only authorized roles can view formula-related content.
- A service restart operation can be performed to confirm that there are no configuration parsing or dependency loading errors in the system.
The values provided are common starting points and should be measured against the reader's own samples.