This page is written for FastGPT v4.17.0, and supports private deployment requirements for semiconductor wafer fabrication scenarios.
What the Data for This Category Looks Like
Daily documents for this category include process flow cards, batch operation records, equipment recipes, SPC charts, exception handling sheets, defect maps, and yield reports. Data sources cover MES systems, equipment FDC, defect detection and metrology equipment, wafer maps, process recipe libraries, and experimental data.
The system updates data by wafer batch and process node. Update frequencies include per-wafer-batch output, real-time parameter reporting, and daily yield summaries.
Fields include wafer batch numbers, nanoscale process parameters, defect spatial coordinates, and yield proportions. Some unstructured data consists of visual images of wafer surface defects.
Constraints Imposed on Deployment and Upgrade
Because data sources span multiple industrial systems with wide update frequency ranges, deployments must meet configuration requirements for cross-system data synchronization to ensure consistency of multi-source data.
Complex process correlations and nanoscale parameters require sufficient resources to be reserved for database connection pools to avoid timeouts during high-concurrency writes and queries.
Unstructured data such as wafer maps and defect maps requires configuration of dedicated storage paths to ensure reading efficiency and security across cleanroom nodes.
Additionally, intranet isolation requirements for wafer fabrication scenarios require adjustments to system access domain names and permission configurations to prevent unauthorized external access.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 200–300 connections | Adapts to high-concurrency read/write requirements for wafer fabrication data, matches connection pressure from multi-source systems, and avoids timeout errors |
MONGODB_URI | mongodb://[account]:[password]@[MES database address]:27017/fastgpt_wafers?authSource=admin | Connects to batch operation records and yield data from wafer fab MES systems, specifies a dedicated database instance to isolate business data |
REDIS_URL | redis://:[cache password]@[cache server address]:6379/0 | Caches real-time updated process parameters and defect detection results to ensure low-latency reading of business data |
STORAGE_S3_ENDPOINT | http://[object storage address]:9000 | Stores unstructured data such as wafer maps and defect maps, adapts to reading requirements across cleanroom nodes |
FILE_TOKEN_KEY | Generate a high-strength random key based on actual testing, set validity period to 7200 seconds | Ensures download permissions for unstructured data and prevents leakage of sensitive wafer fabrication business data |
ROOT_KEY | Generate a random string based on actual testing | Ensures access security for system background management interfaces, complies with permission control requirements for industrial scenarios |
Three Common Mistakes
- Symptom: After modifying the
config.jsonconfiguration, the system model list does not update. Entering the container and checking the/app/data/configdirectory shows no corresponding configuration items. Cause: The file was not modified synchronously in the docker-compose mounted configuration directory, or the mount path was configured incorrectly, causing the system to fail to load the latest configuration. - Symptom: An error is triggered when clicking the knowledge base, with a 500 status code returned. Cause: The
MONGODB_URIorPG_URLparameters were not configured correctly, causing a database connection failure and inability to read wafer fabrication-related business data. - Symptom: No output content is displayed after the question and answer session, and the message is only visible in the details page. Cause:
DB_MAX_LINKwas not adjusted to a value adapted to high concurrency, causing multi-source data reading timeouts and inability to complete splicing and display of response content.
How to Confirm Configuration Completion
- Run the docker-compose ps command to confirm that all service containers are in a running state, with no abnormal restarts or exit records.
- Access the internal domain name configured in
FE_DOMAIN, log in to the database connection test page in the system background, verify the connectivity of theMONGODB_URIandREDIS_URLparameters, and confirm there are no connection errors. - Upload a test wafer map file, use the system's file storage function to verify that the storage path configured in
STORAGE_S3_ENDPOINTcan be read and written normally, with no upload failures or reading exceptions. - Initiate a query for wafer defect analysis, confirm that the response content can be displayed normally, with no timeouts or empty output, and that corresponding business data can be associated.
The values provided are common starting points and should be measured against the reader's own samples.