What the data for this category looks like
Data for this category comes from registered process documents, raw material batch ledgers, DCS control systems, weighing systems, LIMS laboratory management systems, chromatography data systems, and QMS quality management systems. Daily documents include batch production records, material balance sheets, reaction and purification step records, cleaning validation records, impurity profile reports, deviation and change documents. Data updates are triggered per production batch and per process step. Data fields include batch number, reaction temperature, impurity content, solvent residue, yield value, and corresponding testing units and process step identifiers.
What constraints do these characteristics impose on deployment and upgrade
The data characteristics of this category impose multiple constraints on deployment and upgrade. Multi-source heterogeneous data sources require adaptation to interface specifications of different systems. Data synchronization link configuration for all systems must be completed before deployment. High-frequency updates per batch and step require high-throughput storage and indexing capabilities to avoid data delays that impact real-time traceability. Significant differences in structural complexity across different documents require support for mixed parsing and field extraction from unstructured text and structured tables. Full-process traceability requirements require pre-configured association rules across reaction steps and intermediate transfer sequences. This ensures complete coverage of the entire process chain during defect analysis.
How to set the configurations
These configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Approach | Rationale |
|---|---|---|
| `MONGODB_URI | Configure as the dedicated connection string for the active pharmaceutical ingredient data cluster, including read-write permission accounts and port | The data volume of this category grows continuously with production batches. A dedicated cluster isolates business data from traffic from other scenarios, ensuring query efficiency |
| `REDIS_URL | Configure as a local cache cluster, set cache duration to 300–600 seconds | High-frequency batch data queries require caching intermediate results to shorten traceability response times |
| `STORAGES3ENDPOINT | Configure as a compliant object storage endpoint, classify storage buckets by process step | Documents such as impurity profile reports and batch records require unified storage and permission control. Classified storage enables fast retrieval |
| `FILETOKENKEY | Configure as a key format aligned with the LIMS system | Used to verify access permissions for documents exported from chromatography data systems and LIMS, ensuring data security |
| `FE_DOMAIN | Configure as the enterprise’s internal exclusive domain name | Restrict access to internal employees, preventing external unauthorized users from accessing production-related sensitive data |
| `DBMAXLINK | Set to 20–30 | Adapt to concurrent requirements of traceability queries initiated by multiple departments, preventing connection exhaustion |
Three common errors
- After importing process documents, the knowledge base fields are empty, requiring manual category selection. The cause is that
FILE_TOKEN_KEYis not configured with a key format aligned with the LIMS system, preventing the system from automatically extracting batch numbers and process step fields from documents. - Integrated FLUX-class plugins fail to trigger retrieval normally, even when all parameters are configured correctly. The cause is that
REDIS_URLis not configured as the cache link for plugin calls, leading to plugin request timeouts or inability to retrieve context data. - After logging into the system, the interface and knowledge base display in English. The cause is that
FE_DOMAINis not configured to bind the enterprise’s internal language environment variables, causing the system to load the English localization configuration by default.
How to confirm successful configuration
- Run a batch traceability test. Enter a known raw material batch number, and verify that the system automatically pulls corresponding batch production records, impurity profile reports, and process step data.
- Initiate multiple concurrent traceability query requests, and verify that the system returns results stably, with no connection exhaustion or timeout errors.
- Upload an impurity profile report, and verify that the system automatically identifies batch numbers and impurity content fields in the report, and associates them with the corresponding process step.
- Check the system’s storage bucket classification, and confirm that each process document has been categorized according to preset rules, and can be quickly retrieved via keywords.
The values provided are common starting points and should be measured against the reader's own samples.