What this category of data looks like
Data for this category primarily originates from daily business systems of wafer fab process integration, equipment, yield, and other departments. This includes MES production execution data, real-time equipment FDC parameters, wafer maps produced by defect detection and metrology equipment, process recipe library documents, batch operation records, SPC control charts, exception handling sheets, defect atlases, and yield reports. Data update frequencies cover each wafer lot transfer, real-time parameter reporting, and daily summary statistics. Each wafer lot corresponds to process route records and rework path records for hundreds of processes. Fields include nanoscale process parameters, batch numbers, yield values, and defect coordinates. Process and batch identifiers must be associated to ensure data consistency.
What constraints these characteristics impose during deployment and upgrade
Data for this category is multi-source, heterogeneous, and has uneven update frequencies. During deployment, API interfaces of multiple heterogeneous systems including MES and equipment FDC systems must be connected. Data synchronization timing and permission configurations must be planned in advance. The coupling of ultra-long process chains and nanoscale parameters leads to large per-lot document volumes and high field precision requirements. Sufficient storage and index resources must be reserved during deployment. Business continuity requirements in cleanroom environments mean upgrade processes must support gray releases and rapid rollbacks. This prevents system outages from impacting wafer production lots. Limited rework windows require that deployment and upgrade processes do not occupy excessive business processing time. Data loading and retrieval response speeds must be optimized.
Configuration Recommendations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
DB_MAX_LINK | 200–300 | Wafer fabrication data has high-concurrency batch record queries and import requests. This value prevents database connection exhaustion. |
MONGODB_URI | Wafer fabrication dedicated database cluster address with read-write separation suffix | Both historical data such as yield reports and defect atlases, and real-time updated batch operation records must be read. Read-write separation improves retrieval efficiency. |
REDIS_URL | redis://:custom password@cluster address:6379/0 calibrated based on actual testing | Used to cache real-time reported equipment FDC parameters and SPC chart data. Must adapt to high-frequency update business rhythms. |
STORAGE_S3_ENDPOINT | Object storage endpoint that complies with industry data storage specifications | Wafer fabrication documents require high-reliability storage. This configuration ensures persistence of process route sheets, exception handling sheets, and other related documents. |
FE_DOMAIN | Intranet or public network domain where the deployment is located | Used to generate external links and file access tokens for bid documents, and adapts to the internal office network environment of wafer fabs. |
Three Common Misconfigurations
- Symptom: Docker containers remain in the starting state for an extended period, and the frontend interface displays a loading spinner. Cause:
MONGODB_URIis not correctly configured to point to the wafer fabrication dedicated database cluster, resulting in database connection timeout. - Symptom: After batch importing process route sheet documents, search results are empty. Cause: No index mapping for wafer fabrication data is added to the docker-compose configuration file, so no valid document search index is established.
- Symptom: The mongoose version cannot be viewed via the
npm listcommand, and only version information can be obtained from the package.json file. Cause: Complete npm dependency packages are not installed in the FastGPT container during deployment. Only the version is specified via the lock file, so local dependencies cannot be queried directly.
How to Confirm Proper Configuration
- Run the
docker logs fastgpt-webcommand to view the startup logs of the FastGPT v4.17.0 container. Confirm there are no database connection timeout or storage endpoint errors. - Upload a single wafer lot's process route sheet document, run a keyword search test, and confirm matching document results are returned.
- Log in to the object storage console, confirm that wafer fabrication-related documents have been successfully synchronized to the specified
STORAGE_S3_ENDPOINTstorage path. - View the Redis monitoring panel, confirm that cache connection counts are within a stable range, with no abnormal disconnections or overflow issues.
The values provided are common starting points and should be measured against the reader's own samples.