What the Data for This Use Case Looks Like
Petroleum refining scenario data covers daily documents from oil refinery production operations, process, scheduling, equipment, and laboratory departments. These include operating procedures, production plans, material balances, distillation outlet quality reports, interlock ledgers, shift handover records, and change management documents. Data sources include DCS/APC, LIMS, tank farm metering, crude oil evaluation, equipment management systems, and energy systems. Update frequencies cover real-time, shift-based, and daily. Documents are categorized by production process and equipment management. Fields include equipment numbers, operating parameters, test indicators, maintenance records, and more. Units use industrial standard units such as pressure, temperature, and flow. Individual document sizes range from tens of KB to several MB.
Constraints Imposed by These Characteristics During Deployment and Upgrade
Large differences in update rhythms across multiple data sources require adapting to real-time, shift-based, and daily synchronization frequencies during deployment. This avoids data delays or database connection overload. Differences in structure and fields across multiple document types require configuring targeted text processing rules during deployment. This adapts to the formats and units of different documents. High-frequency updates of real-time data require reserving sufficient system resources to handle real-time requests. Stability of real-time data access must be verified during upgrades. Documents related to equipment inspection and maintenance have strong correlations. Association retrieval logic must be configured during deployment. This ensures maintenance Q&A can quickly match historical records and operating procedures for corresponding equipment, improving retrieval accuracy.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 100–150 connections | Adapts to real-time, shift-based, and daily update frequencies of multi-source data, avoiding database connection overload |
MONGODB_URI | Points to an independently deployed high-availability replica set address | Stores unstructured documents such as equipment ledgers and maintenance records, ensuring stable read/write performance for multiple data types |
REDIS_URL | Configured as a clustered address with persistence | Caches real-time equipment parameters and retrieval results, matching the rhythm of real-time updated data |
STORAGE_S3_ENDPOINT | Connects to the internal network address of the enterprise’s internal object storage service | Stores large-volume documents such as crude oil evaluation reports and interlock ledgers, adapting to large-file storage requirements in industrial scenarios |
ROOT_KEY | Generated according to internal enterprise permission rules | Ensures access permission security after system deployment, complying with information security requirements of oil refining enterprises |
Three Common Misconfigurations
- Issue: Returns referenced content but the reference switch is not enabled. Cause: After upgrading to FastGPT v4.17.0, the configuration items for the default reference logic were not updated synchronously, resulting in mismatched parameters and version.
- Issue: Tool call judgment logic malfunctions, and some models cannot call preset tools. Cause: The permission verification rules corresponding to
ROOT_KEYwere not configured during deployment, resulting in restricted tool call permissions. - Issue: Ollama access fails for non-Docker deployments. Cause: Network intercommunication rules between
FE_DOMAINand internal services were not correctly configured during deployment, resulting in cross-service requests being blocked.
How to Verify Proper Configuration
- Run a database connection test to verify whether the connection number set by
DB_MAX_LINKcan support synchronization requests for multi-source data. - Submit different types of petroleum refining equipment documents to check whether retrieval results correctly match the fields and units of corresponding documents.
- Trigger a real-time equipment parameter retrieval request to confirm that response delay meets the real-time requirements of the business scenario.
- Verify the tool call logic to confirm that the configured permission rules can properly block unauthorized access requests.
The values provided are common starting points and should be measured against the reader's own samples.