What the data for this category looks like
Data for this category comes from design BOMs, silicon steel sheet and insulating material procurement data, vacuum drying equipment logs, insulating oil test reports, high-voltage test station records, and documents for individual products including winding assembly, vacuum drying, and nameplate parameters. Data is updated per product or batch of raw materials. Documents include clear physical parameter fields, such as insulation thickness (unit: mm), drying temperature (unit: ℃), and test voltage (unit: kV). Each document is tied to a unique product serial number. Cross-check data from multiple links must be associated to complete the review.
What constraints these characteristics impose on deployment and upgrade
The batch-updated data characteristics for individual products require deployment setups to support low-latency, small-batch data writes, to avoid excessive resource usage from batch indexing tasks. The need for cross-link association across multiple document types requires pre-configured field matching rules during deployment, to adapt to the field structures of different documents such as insulating oil reports and test records. The unique product serial number as the core association identifier requires fixed field verification logic to prevent data binding errors. The storage requirement for high-precision physical parameters requires reserving sufficient storage and memory resources during deployment, to support parsing and indexing of large-value fields.
How to set the configurations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://fastgpt:xxxx@localhost:27017/fastgpt?authSource=admin | Adapts to the associated storage of multi-link documents, supports quick querying of cross-check data by product serial number |
REDIS_URL | redis://:xxxx@localhost:6379/1 | Caches frequently read insulating parameters and test results, reduces database concurrency pressure |
DB_MAX_LINK | 25 | Adapts to the scenario of frequent updates to multiple documents per product, avoids connection pool resource exhaustion |
FE_DOMAIN | https://transformer-manufacture.example.com | Configures a compliant access domain name, meets the secure access requirements of industrial scenarios |
STORAGE_S3_ENDPOINT | http://192.168.1.100:9000 | Stores large-volume documents such as insulating oil test reports and withstand voltage test records, adapts to the document storage needs of this category |
Three common mistakes
- Phenomenon: Server disk space is rapidly exhausted after deployment. Run the
docker volume lscommand to identify a large number of uncleaned temporary parsing files. Cause: No regular cleanup task is configured. Parsing of large-volume high-voltage test reports for this category generates a large amount of temporary cache data. - Phenomenon: The frontend page cannot be accessed via the custom domain name, and the browser returns an insecure connection prompt. Cause: The binding relationship between
FE_DOMAINand the SSL certificate is not correctly configured, and the container internal port is not mapped to port 443. - Phenomenon: Index construction fails, and the console returns an error indicating that the specified embedding model cannot be found. Cause: The official embedding model identifier supported by FastGPT v4.17.0 is not used, and a non-standard model name parameter is incorrectly configured.
How to confirm the configuration is correct
- Run the database connection test command to verify that the
MONGODB_URIandREDIS_URLconfigurations can establish normal connections. - Upload an insulating oil report document, check whether it can be normally stored to the specified object storage path, to verify the validity of the
STORAGE_S3_ENDPOINTconfiguration. - Initiate a small-batch document indexing task, check whether the database connection count is within the range set by
DB_MAX_LINK. - Access the configured
FE_DOMAIN, verify that the frontend page loads normally and there are no security certificate errors.
The values provided here are common starting points, and should be evaluated against applicable deployment samples.