What the data for this category looks like
Data comes from hot metal and scrap charging data, L2 process control systems, laboratory composition reports, MES systems, energy medium monitoring systems, and equipment status collection nodes. Update frequencies cover three categories: per-heat, real-time, and shift-based. Daily documents include heat plans, batching lists, converter smelting records, composition analysis reports, continuous casting process parameter records, and slab quality judgment sheets. The data structure includes structured numerical fields and unstructured scanned documents. Field units include industrial standard units such as temperature, component proportion, and flow rate. The process is high-temperature continuous operation, with tight coupling between heats and casting sequences, and data collection logic cannot be adjusted arbitrarily by shutting down the process.
What constraints do these characteristics impose on deployment and upgrade
Multi-source heterogeneous data sources require that the deployment phase supports integration with multiple system interfaces, and avoids hard-coded data source configurations. Real-time and high-frequency updated data requires database connection pools to adapt to high-concurrency writes, and data writes must not be interrupted during upgrades. The tight coupling between heats and casting sequences requires that the time window for ledger summarization strictly matches the production rhythm, and data aggregation logic cannot be adjusted arbitrarily during upgrades. The high-temperature continuous process requires that upgrade operations use a gray release model to avoid production interruptions caused by full upgrades. The storage requirement for unstructured documents requires configuring a dedicated object storage service, separated from structured data storage.
How to set configurations
| Configuration Item | Recommended Setting | Rationale for This Setting |
|---|---|---|
DB_MAX_LINK | 20–30 connections | Adapts to the concurrency requirements of per-heat real-time writes, and avoids database connection exhaustion |
PG_URL | MES ledger database connection string containing port and authentication information | Stores structured converter smelting records, composition reports, and continuous casting process data |
REDIS_URL | Configure cache expiration time to 600 seconds | Caches shift-based scheduling coordination data, balancing real-time performance and resource usage |
STORAGE_S3_ENDPOINT | Bind industrial object storage nodes with bandwidth no less than 100 MB/s | Stores unstructured scanned documents such as heat plans and slab judgment sheets |
ROOT_KEY | Generate a dedicated key according to the enterprise IAM system | Controls highest-privilege system access, complying with process industry data security specifications |
FE_DOMAIN | Configure as the enterprise internal secondary domain name, port 8080 | Provides access entry for ledger query and report generation |
Three common mistakes
- Symptom: All ledger historical data and system configuration stored in the container are lost after restarting the server. Cause: The database, cache, and document storage directories were not mounted to the host's persistent storage volumes, resulting in data being destroyed along with the container instance after restart.
- Symptom: Connection timeout errors frequently occur when executing multi-heat ledger summarization tasks, with 504 status codes returned in logs. Cause: The
DB_MAX_LINKconfiguration value is too low, and the concurrent connection count exceeds the limit, causing request queuing and timeout. - Symptom: Some time periods of energy medium consumption data are missing in the generated continuous casting process report. Cause: Not all data source integration interfaces were configured, and only data from some production nodes was synchronized.
How to confirm proper configuration
- Execute the
docker pscommand, confirm that FastGPT v4.17.0 related containers and mounted persistent storage volumes are in running state. - Initiate a single-heat ledger data synchronization task, check whether complete structured records for the corresponding heat are written to the target database.
- Access the domain name configured in
FE_DOMAIN, log in to the system and view the historical ledger list, confirm that all types of documents can be loaded and previewed normally. - Trigger a monthly production report generation task, check whether the output report contains complete content including heat plans, smelting records, continuous casting process parameters, and quality judgment results.
The values provided are common starting points and should be measured against the reader's own samples.