What data for this category looks like
Daily documents for this category include bill of materials (BOM), assembly work instructions, refrigerant charging records, electrical safety test reports, performance sampling inspection reports, rework records, and more. Data sources cover ERP/MES systems, component batch ledgers, tightening and charging equipment, safety compliance testers, performance laboratory equipment, and after-sales fault databases. Data updates follow per-unit product, shift production batch, or sampling inspection schedules. Fields include component batch numbers, process parameter values, test results, rework reasons, and more. Some documents are structured process ledgers, while others are unstructured test reports and rework records.
What constraints these characteristics impose on deployment and upgrade
Deployments must adapt to multi-source, high-frequency inspection data from multiple types of equipment to avoid data synchronization delays. Systems must support complex nested parsing rules for multiple document structure types, to adapt to process records and test reports in different formats. Storage and cache configurations must provide high availability to handle full-process records for individual products and concurrent writes of batch sampling inspection data, preventing data loss or upload failures. Systems must accurately match documents and process data for multiple models in mixed-line assembly production scenarios, to avoid misassembly risks. Upgrades must use a gray release strategy due to severe potential error consequences, to ensure stability of the production environment.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
MONGODB_URI | Configure as a high-availability connection string supporting replica sets | Adapt to the storage requirements of high-frequency written inspection records from multiple devices, ensure data persistence and high-concurrency access |
REDIS_URL | Configure as a clustered connection address with persistence, memory quota no less than 16 GB | Respond to high-frequency process status caching and session synchronization, meet real-time requirements for shift updates |
STORAGE_S3_ENDPOINT | Configure as a local object storage endpoint compatible with the S3 protocol | Store large-capacity unstructured documents such as performance test reports and rework records, adapt to document archiving requirements for multiple model types |
DB_MAX_LINK | Set to 200–300 | Match the upper limit of database connections for concurrent access from multiple devices, prevent record upload failures caused by exhausted connections |
FILE_TOKEN_KEY | Configure as a randomly generated 32-bit encrypted string | Ensure permission verification during inspection document uploads, prevent unauthorized document tampering and leakage |
ROOT_KEY | Configure as a complex random string and rotate it regularly | Restrict administrator operation permissions, avoid mismodification of production environment configurations that disrupt production processes |
Three common mistakes
- An
embed ding erroroccurs when uploading a BOM configuration. The cause is failure to adapt to the complex nested structure of multi-model BOMs in complete home appliance manufacturing. The default parsing rule cannot recognize hierarchical component fields. - Search plugins fail to work properly when deploying with docker-compose. The cause is failure to configure
FE_DOMAINto the actual domain name of the production environment, leading to cross-domain requests from plugins being blocked by the gateway. - A
'usage' Keyeerror appears in logs after upgrade. The cause is failure to adapt to the field format of after-sales fault data for home appliances. Context parameters for large model calls are not correctly mapped to fields in after-sales records, leading to missing expected usage-related fields in returned results.
How to confirm configurations are correct
- Upload an air conditioner configuration BOM document, check that fields such as component batches and assembly processes are correctly extracted after parsing, with no parsing failure prompts.
- Send concurrent inspection record upload requests from multiple devices, verify that database connection counts do not exceed the
DB_MAX_LINKsetting value, with no connection timeout errors. - Modify
FILE_TOKEN_KEY, then attempt to upload a document without a valid token, confirm that the request is blocked, and verify that the permission verification function works. - Deploy the new version of FastGPT v4.17.0 on a non-production node, verify that document parsing, storage, and query functions work normally before performing a full upgrade.
The values provided here are common starting points, and should be evaluated against the reader's own samples.