What the data for this category includes
Data for this category primarily comes from cell capacity grading data, MES systems, welding equipment, torque tools, BMS software version records, and EOL test benches. Daily documents include grouping rules, BOM lists, welding and torque operation records, BMS flashing logs, insulation and voltage withstand reports, EOL test reports, and system configuration lists. Data is updated once per produced battery pack or energy storage cabinet, and synchronized each time the BMS software version changes. Data fields cover cell parameters, welding torque values, insulation resistance values, BMS configuration items, and more. Units include milliampere hours, newton meters, megaohms, and others.
What constraints these characteristics impose during deployment and upgrade
This category’s data is associated with production lines on a per-device basis and updates frequently. Deployments must support isolating data storage by production line to avoid mixing inspection records from different lines. Frequent version changes and independent data requirements for single devices require configuring quickly switchable knowledge base data sources and version synchronization mechanisms. This ensures upgrade processes do not interrupt real-time data collection. Data access from multiple types of heterogeneous devices requires deploying data source adapters compatible with different protocols. These adapters must adapt to the data formats of welding equipment, torque tools, and other devices. Additionally, compliance requirements for inspection records ensure that documents stored in the knowledge base fully match the fields of real-time collected data. This avoids audit failures caused by missing or mismatched fields. Upgrade processes must retain historical versions of inspection records to facilitate traceability and audits.
Configuration guidelines
This document is compatible with FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 100–150 connections | Meets concurrent access demands for thousands of daily inspection records per production line, and prevents database connection exhaustion |
MONGODB_URI | Exclusive database connection string for the target production line | Enables data isolation for inspection records across production lines, and meets compliance traceability requirements |
REDIS_URL | Intranet address of a local cache instance | Accelerates caching and querying of real-time inspection data, and reduces load on the primary database |
ROOT_KEY | Randomly generated 32-character string | Secures access to production line backend configurations, and prevents cross-production-line permission leaks |
STORAGE_S3_ENDPOINT | Intranet domain name of the enterprise’s internal object storage | Stores large documents such as insulation and voltage withstand reports and EOL test videos, and optimizes storage costs |
Three common configuration mistakes
- Scenario: After an upgrade, the knowledge base can retrieve documents normally, but no content is output when calling the large model. Cause: The
FILE_TOKEN_KEYconfiguration was not synchronized during the upgrade, so the model cannot obtain document access permissions. - Scenario: After Docker deployment, the page on port 3000 cannot be accessed, but the port 3001 page opens normally. Cause: The frontend service port was not mapped correctly during deployment, or the firewall did not open access permissions for port 3000.
- Scenario: After modifying the FastGPT configuration file inside a Docker container, the configuration does not take effect. Cause: The configuration file directory was not mounted when starting the container, or the modified configuration item does not correspond to the startup parameter’s environment variable.
How to verify successful configuration
- Run a connection count test for
DB_MAX_LINK. Simulate production line concurrent requests to confirm that the database connection count does not reach the configured limit. - Call the knowledge base retrieval interface, pass the production line-exclusive
MONGODB_URIparameter, and confirm that inspection record documents for the target production line are returned. - Restart the FastGPT service, then check backend logs to confirm that configuration items such as
REDIS_URLandROOT_KEYloaded normally, with no error messages. - Attempt to modify
ROOT_KEYand restart the service. Confirm that backend login requires the new key to verify that the configuration takes effect.
The values provided here are common starting points, and should be evaluated against the reader’s own samples.