This page configuration applies to FastGPT v4.17.0.
What this category’s data looks like
Data for ready-mixed concrete production comes from commercial concrete batching plants, project laboratories, and quality and scheduling departments. Daily processed documents include mix proportion notices, opening identification forms, production and transport slips, raw material inspection records, test block retention records, and strength test reports. Data sources cover engineering orders, sand, gravel, cement, and admixture inspection reports, moisture content measurement records, batching plant control system data, and laboratory ledgers. Updates follow the rhythm of each production batch per vehicle, each raw material batch, and daily ledger updates. Data fields include project number, raw material batch, moisture content value, pouring location, strength grade, transport duration, and other related items.
What constraints these characteristics impose on deployment and upgrade
This category’s data is updated frequently by vehicle, batch, and day. During deployment, adapt to high-concurrency real-time data synchronization requirements to avoid write failures caused by insufficient connection pools. Multiple heterogeneous document formats are inconsistent. During upgrade, maintain compatibility with field extraction logic for different templates to ensure complete information in tender documents. Data sources include on-site control systems and laboratory ledgers. During deployment, grant cross-system data pull permissions to avoid data silos. Tender quotations must be linked to accurate raw material costs and transport timelines. During upgrade, verify the stability of cross-module data linkage to prevent quotation deviations.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 20–30 connections | This category’s data is updated frequently. Each vehicle or batch of data triggers a database write, so the connection pool must be sufficient to support concurrent requests |
REDIS_URL | redis://:custom password@redis host address:6379/0 | Cache frequently queried raw material prices and engineering order information to reduce database load. Configure an access password to secure cached data |
STORAGE_S3_ENDPOINT | Fill in according to the enterprise's self-built object storage address | Documents such as test reports and transport slips for this category require unified storage. Adapt to the endpoint address of the enterprise’s existing storage service |
FE_DOMAIN | https://internal enterprise tender system domain name | Sync quotation documents generated by FastGPT to the internal tender platform. Configure a valid front-end domain name to enable cross-domain access |
ROOT_KEY | Randomly generated 32-bit string | Secure access to the backend management interface and prevent unauthorized configuration modifications |
FILE_TOKEN_KEY | Randomly generated 16-bit string | Encrypt uploaded document attachments to prevent unauthorized access to sensitive files such as scanned copies |
Three common mistakes
- Phenomenon: Image pull times out when running
docker-compose up, and logs indicate a failed connection to the GitHub mirror source. Cause: No domestic mirror source is configured to speed up pulls. This category’s business images depend on numerous basic components, resulting in long pull times. - Phenomenon: A deserialization error is returned when calling the knowledge base retrieval interface, and the front-end page cannot display results. Cause: Incorrect
FILE_TOKEN_KEYconfiguration leads to failed parsing of document attachments. Test reports for this category include binary format scanned copies, and parsing exceptions trigger field deserialization failures. - Phenomenon: After deploying multiple replicas in K8s, some nodes cannot sync the latest raw material data. Cause: No shared cache is configured for
REDIS_URL, and local caches of each replica are inconsistent, leading to unsynced data updates.
How to confirm successful configuration
- Execute the
docker-compose pullcommand, and confirm that all business images can be pulled within 5 minutes. - Log in to the FastGPT backend management interface, upload a ready-mixed concrete mix proportion notice, and confirm that the document can be parsed normally and extract fields such as project number and raw material batch.
- Start 2 K8s replicas, submit a new raw material inspection data record, and confirm that both replicas can retrieve the data.
- Call the interface for generating tender documents, and confirm that the returned document includes correct fields such as raw material costs and pouring timelines.
The values provided are common starting points and should be measured against the reader's own samples.