Reference only; verify against the target FastGPT release and deployment documentation.
What the data for this use case looks like
Data for this use case comes from the China Center for Disease Control and Prevention Information System, hospital HIS systems, infectious disease reporting systems, and laboratory testing systems. The update frequency is daily.
Daily processed documents include infectious disease report cards, epidemic analysis reports, case correction records, clustered outbreak reports, and epidemiological investigation reports. These cover two format types: structured form data and unstructured analysis text.
Data fields include case diagnosis information, reporting time limits, correction flags, and other items. Configuration must strictly match relevant requirements for statutory infectious disease reporting. This data is maintained daily by disease control institutions, public health departments of medical institutions, disease control divisions of health and health commissions, and other entities.
What constraints do these characteristics impose during deployment and upgrade
Daily high-frequency updated data requires deployment to support incremental synchronization and scheduled verification tasks, to avoid resource waste from full pull operations.
Multi-source heterogeneous data sources require configuration of multiple types of database connection parameters, to ensure normal reading of case data from different systems.
Diversity of document formats requires that the chunking strategy during deployment can adapt to structured form fields and unstructured analysis text, ensuring accuracy of index generation.
Rigor of statutory reporting requires that the deployed system retains complete operation logs, and strengthens the abnormal alert mechanism, to avoid issues such as underreporting or delayed reporting affecting epidemic monitoring efficiency.
How to set configurations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
DB_MAX_LINK | 10–15 connections | Adapts to daily high-frequency case data query and synchronization requests, avoids connection exhaustion |
PG_URL | Intranet-only database connection string with read-only permission | Connects to structured case data from the disease control information system, ensures data access security |
REDIS_URL | Intranet cache connection with 600-second cache expiration time | Caches high-frequency queried epidemic summary data, reduces access pressure on the core database |
STORAGE_S3_ENDPOINT | Intranet object storage endpoint | Stores unstructured documents such as epidemiological investigation reports and correction records, complies with intranet deployment security requirements |
MONGODB_URI | Intranet MongoDB connection string for storing audit logs | Retains audit and correction logs for infectious disease reports, meets traceability and compliance requirements |
Three common misconfigurations
- Phenomenon: The
EMFILE: too many open fileserror appears when building a Docker image. Cause: The container's default file handle limit is insufficient to support batch chunking and index generation operations for infectious disease report documents. - Phenomenon: No valid results are returned when importing text using the bge-large vector model after intranet deployment with docker-compose. Cause:
REDIS_URLis not correctly configured as a temporary cache, resulting in inability to persist segmented document data. - Phenomenon: Generated epidemic analysis reports lack case correction record fields. Cause: Insufficient database permissions configured for
PG_URL, unable to read correction data sources from the disease control information system.
How to confirm the configuration is successful
- View running logs of all service containers, confirm there are no connection timeout, permission error, or cache exception prompts.
- Upload a single test infectious disease report card document, verify that chunking processing and index generation processes complete normally.
- Call the intranet interface to query newly added case data for the current day, confirm that returned fields are complete and meet business requirements.
- Test vector model calls in the intranet environment, confirm that index results matching the input document can be generated.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.