What data for this category looks like
This category covers end-to-end operations for carbonated drink, juice, tea, functional beverage, and packaged drinking water factories. Daily documents include formula sheets, water treatment records, blending records, sterilization or aseptic filling records, packaging material inspection records, and finished product inspection records. Data sources include water source and raw material testing systems, formula management systems, online sugar and acidity monitoring equipment, filling equipment, visual inspection equipment, and LIMS systems. Data updates happen per production batch, per shift inspection, and via continuous real-time collection. Document fields include concentrate feeding ratio, additive dosage, sterilization temperature, filling liquid level parameters, and corresponding testing units.
Constraints during deployment and upgrade
This category has large volumes of continuously collected real-time data, mostly structured production and testing data. During deployment, the system must support high-concurrency document reading and index construction. The ability to switch between multiple batches and product categories requires the system to support dynamic switching of process document templates. During upgrades, the system must be compatible with differentiated field configurations across different production lines. Abnormal data from water treatment, sterilization, and other links must be associated with reports in real time. During deployment, configure time-sensitive thresholds for data synchronization to avoid delayed abnormal handling. Document formats vary across different production lines. During upgrades, retain existing field mapping rules to avoid overwriting historical production data.
Configuration guidelines
This document applies to FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
STORAGE_S3_ENDPOINT | Match the access address of the production line’s object storage, such as http://192.168.1.100:9000 | Most production documents and testing data for this category are stored in on-site object storage. Ensure the storage address aligns with existing systems |
REDIS_URL | Configure as the address of the production line’s dedicated cache instance, such as redis://:password@192.168.1.101:6379/0 | Real-time collected filling and sugar content data must be cached. A dedicated instance avoids conflicts with other business caches |
DB_MAX_LINK | Set to 200 | This category has a high data update frequency. 200 connections cover peak concurrency when multiple production lines read documents and abnormal data simultaneously |
MONGODB_URI | Bind to the database instance address synchronized with the production line’s LIMS system | Store structured documents such as formula sheets and inspection records. Align with existing data links |
FILE_TOKEN_KEY | Generate a 32-bit random string as the secret key | Used to encrypt access tokens for process documents. Ensure only authorized production line equipment can read sensitive production data |
Three common configuration errors
- Phenomenon: When creating a beverage category knowledge base with an instance deployed via Sealos, the process gets stuck at the index construction step. Logs show storage connection timeout. Cause:
STORAGE_S3_ENDPOINTis not correctly configured to point to the production line’s dedicated object storage, preventing normal reading of process documents. - Phenomenon: After switching from an application with mandatory global variables to an application without global variables, mandatory variable validation errors still trigger, and the application interface cannot be accessed. Cause: Front-end cached old global variable configurations were not cleared during deployment or upgrade, and front-end resources were not force-refreshed.
- Phenomenon: After modifying configurations when starting an instance via Docker, the instance cannot read filling equipment data from the production line. Cause: The Docker container was not restarted to load the new configuration, or the configuration file was not mounted to the corresponding directory in the container.
How to confirm configurations are correct
- Log in to the FastGPT backend, check the connection status of the cache instance corresponding to
REDIS_URL, and confirm there is an active connection count. - Upload a standard beverage formula sheet document to trigger index construction, confirm the index progress bar completes normally with no error prompts.
- Switch to process document templates for different production lines, confirm the system can normally load corresponding fields with no missing or incorrect mappings.
- Call the interface to test the connection limit for
DB_MAX_LINK, confirm no connection overflow errors trigger when multiple production lines read data simultaneously.
The values given are common starting points and should be measured against the reader's own samples.