What data for this category looks like
Data for this category comes from registered process documents, raw material batch ledgers, DCS systems, weighing recording systems, LIMS, chromatography data systems, and QMS. Daily outputs include batch production records, material balance sheets, reaction and purification step records, cleaning validation records, impurity profile reports, deviation and change documents. Data is updated per production batch and per process step. Fields include batch number, process step identifier, operation timestamp, raw material feeding quantity, reaction yield value, impurity content value, solvent residue value, and similar fields. No preset statistical percentage fields are included.
Constraints on database and operations maintenance from these characteristics
Multi-source heterogeneous data sources require the database to support flexible field mapping and format compatibility, to avoid data conflicts between different systems. The characteristic of frequent updates per batch step generates continuous small-volume write requests. Concurrent connection counts must be limited to prevent database blocking. Highly compliant document content requires the database to retain full-chain audit trails, so detailed write logs must be enabled. Professional chemical parameter precision requirements are high, so appropriate data storage precision must be configured to avoid parameter distortion. Data volume grows linearly with production scale, so storage capacity and hot/cold data separation strategies must be planned reasonably.
How to configure settings
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://fastgpt-db:27017/med_pharma_batch?retryWrites=true&w=majority | Adapt to unstructured storage needs for multi-source chemical data, support batch writing and querying |
DB_MAX_LINK | 200–300 connections | Match the concurrent scenario of frequent small-volume writes per process step, avoid exhausting database connections |
REDIS_URL | redis://redis-cache:6379/1 | Cache impurity profile and yield data of popular batches, reduce query pressure on the primary database |
PG_URL | postgresql://postgres:fastgpt@pg-db:5432/med_pharma_meta | Store metadata and change records of compliant documents, implement hot/cold data separation storage |
Three common configuration errors
- Symptom: Running
npm list mongooseinside the FastGPT container returns no output, but mongoose is listed as a dependency in package.json. Cause: Dependencies are only installed in the project directory within the container, and project-specific environment variables are not activated, causing the global npm command to fail to recognize the dependency version. - Symptom: Database queries return empty fields or mismatched structures. Cause: Database mapping is not configured according to the field specifications for active pharmaceutical ingredient production data, and core business fields such as batch number and impurity content are not aligned.
- Symptom: Database write timeouts or connection failures. Cause: The
DB_MAX_LINKconfiguration is not adjusted, and concurrent write requests exceed the connection pool limit, leading to exhausted connections that cannot respond to new requests.
How to verify the configuration is complete
- Run the
echo $MONGODB_URIcommand, confirm that the output connection string matches the configuration item, and adjust connection parameters according to production scale. - Check database operation logs, confirm that multi-source data writes have no format errors and match the preset field structure.
- Initiate a simulated batch data write request, confirm that the write is successful and there are no timeout errors, and adjust the timeout threshold based on actual measurements.
- Check the cache service hit rate, confirm that popular batch data is correctly cached, and set the cache validity period according to the data update frequency.
The values provided are common starting points and should be measured against the reader's own samples.