This document applies to FastGPT v4.17.0.
What the data for this category looks like
Data sources for production ledger data in steel rolling and finishing include slab or billet batches, heating furnaces, mill automation control systems, online thickness and width measurement equipment, surface inspection equipment, and MES systems. Data updates occur in real time per coil or per billet, covering rolling plans, heating schedules, rolling specifications, shape and thickness records, surface quality reports, quality certificates and other documents. Data fields include batch number, furnace number, rolling speed, reduction ratio, thickness tolerance, surface defect level and more. Units include millimeters, tons, degrees Celsius, meters per second and others, and full-link traceability at the coil level is required.
What constraints these characteristics impose on deployment and upgrade
The real-time coil-level data, multi-source heterogeneous data sources, and strict traceability requirements for this category impose clear constraints on deployment and upgrade. First, high-frequency real-time updated data requires sufficient connection pool and cache resources to avoid data synchronization delays. Second, data in different formats from MES and online inspection equipment requires configuration for multi-source data access to ensure accuracy of field extraction. Third, coil-level traceability requires strong association between ledger data and full-process records of corresponding batches, so index optimization for associated queries must be configured during deployment. Fourth, errors directly relate to coil downgrading or equipment damage, so data verification and exception alert mechanisms must be added during deployment.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 100–200 connections | Adapts to high-frequency concurrent requests from real-time inspection data and MES data, avoiding connection pool exhaustion |
MONGODB_URI | Configured as the associated database address for storing surface quality reports and quality certificates | Used to store unstructured inspection report and quality certificate attachment metadata |
PG_URL | Configured as the read-only database address of the production MES system | Used to obtain structured ledger data such as rolling plans and heating schedules |
REDIS_URL | Local Redis instance or managed Redis service address, with cache validity set to 300 seconds | Caches real-time thickness and width measurement data to reduce database query pressure |
STORAGE_S3_ENDPOINT | Configured according to the object storage service provider used, for example https://s3.cn-north-1.amazonaws.com.cn | Used to store documents and report attachments generated during the rolling process |
FE_DOMAIN | Configured as the official domain name for official website access | Used to generate external link access addresses for reports, ensuring external parties can normally download attachments |
Three common mistakes
- Symptom: No available channel options appear when accessing the OneAPI management page after deployment. Cause:
ROOT_KEYpermissions are not configured correctly, or the configuration file for the corresponding model channel has not been imported. - Symptom: Shape and thickness data is missing in generated ledger reports. Cause:
PG_URLis not pointed to the read-only database of the MES system, or database connection permissions are insufficient to read structured ledger data. - Symptom: Real-time inspection data update delay exceeds acceptable limits. Cause:
DB_MAX_LINKis set too low, causing the database connection pool to be exhausted and unable to handle high-frequency real-time data requests.
How to confirm configuration completion
- Run the FastGPT built-in database connectivity test script to verify that the connection status of
MONGODB_URI,PG_URL, andREDIS_URLis normal. - Upload a rolling plan document to trigger the ledger summary task, and check whether core fields including batch number, heating schedule, rolling speed and others are extracted correctly.
- Access the report link generated via
FE_DOMAINto confirm that attachments such as quality certificates and surface quality reports load normally. - Check system operation logs to confirm that the current number of connections for
DB_MAX_LINKdoes not reach the set limit, and no connection timeout errors occur.
The values provided are common starting points and should be measured against the reader's own samples.