What this category of data looks like
This category of data mainly comes from the unified budget management system, project declaration materials, fund payment data, business system ledgers, and satisfaction surveys. The update frequency is annual. Daily processed documents include performance target forms, self-evaluation forms, performance evaluation reports, supporting material lists, and rectification reports. Document fields include project number, fund amount, performance indicators, supporting material association identifiers, etc. There is no fixed unified unit, and the corresponding accounting unit must be matched according to the specific project. The data association logic is complex, and consistency and relevance of data from different sources must be ensured.
What constraints do these characteristics impose on the deployment and upgrade phase
The annually updated data of this category has strong batch characteristics, with a large amount of data imported in a single operation. This requires configuring storage connection parameters adapted to large-capacity data during deployment. Business data from multiple sources needs to be connected to multiple external systems such as the unified budget management system, so flexible data source configuration space must be reserved. Diverse document types include structured tables and unstructured reports, so configuration items supporting multi-format file parsing must be set during deployment. The complex performance data association logic requires ensuring the stability of database connections and query efficiency, to avoid task interruptions caused by connection overload.
How to set configurations
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 10–15 connections | Adapts to the high-volume database connection requirements of annual batch imports, and prevents connection pool exhaustion |
REDIS_URL | Internal cache instance address | Used to cache performance indicator association data, adapting to the cache frequency of multi-system data synchronization |
STORAGE_S3_ENDPOINT | Internal unit object storage service address | Used to store large-volume documents such as supporting materials and rectification reports, ensuring internal data access permissions |
MONGODB_URI | Independent dedicated instance connection string | Stores unstructured documents such as performance evaluation reports and self-evaluation forms, isolated from the business database |
PG_URL | Read-only permission address of the unified budget management system | Used to pull basic data such as fund payments and project declarations, ensuring data security |
FE_DOMAIN | Secondary domain name of the internal office system | Adapts to the access permissions of internal employee services, only open to the internal unit network |
Three common mistakes
- An OCI runtime create failed error occurs during Docker startup. The cause is that the deployment environment is not configured with GPU-compatible startup parameters, and some images rely on GPU acceleration environments, leading to startup failure.
- Redis image pull fails, even if a network proxy is configured. The cause is that the new version adds Redis image dependencies, and the old version has no corresponding image packages.
- The MCP tool in the workflow cannot obtain the global variable token. The cause is that the tool access permission of the global variable is not enabled, causing the tool to fail to read context parameters.
How to confirm the configuration is correct
- Execute a batch import test task for performance target forms, and check that the database connection pool connection count does not exceed the
DB_MAX_LINKconfiguration value. - Attempt to pull fund payment data from the unified budget management system, and verify that the read-only permission configured in
PG_URLis working properly. - Upload supporting material documents, and check that the storage path matches the
STORAGE_S3_ENDPOINTconfiguration. - Start the MCP tool test, and verify that global variables can be normally passed to tool parameters.
The values given 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.