Data Profile for This Category
Data for this category originates from ERP, WMS systems, barcode/RFID collection devices, AGV conveyance systems, and weighing equipment. Daily documents include receiving orders, putaway orders, material picking and return orders, inventory transfer orders, inventory checklists, lot ledgers, and similar records. Common data fields include material code, lot number, storage location number, expiration date, inbound/outbound quantity, inspection status, operator ID, and timestamp. Update frequencies include real-time (triggered by barcode scanning), daily (work order picking summaries), and monthly (full inventory counts). Core focus areas are material identity, lot flow direction, and storage location status.
Constraints on Database and Operations
Multi-source real-time writing requires databases to support high-concurrency connections and avoid data packet loss. Unique lot numbers and material codes require joint indexes to speed up associated queries. Monthly inventory count bulk data needs sufficient reserved storage capacity and temporary query windows. Handling discrepancies between ledger records and actual inventory requires multi-table associations across documents. It also requires atomicity of data updates to prevent inconsistencies between ledgers and actual stock levels.
Configuration Settings
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
MONGODB_URI | Cluster address with readPreference=secondaryPreferred | Adapts to mixed loads of real-time document writing and ledger queries, separating read and write traffic |
PG_URL | Dedicated structured storage instance address with connection pooling enabled | Stores structured data requiring complex associations such as lot ledgers and storage location mappings |
DB_MAX_LINK | 80–120 connections | Balances concurrent connection count and server resource usage, avoiding connection exhaustion |
REDIS_URL | Local cache instance address with key expiration time set to 300 seconds | Caches frequently queried lot expiration dates and storage location status to reduce database query pressure |
| Bulk import threshold | 500 records per batch | Adapts to bulk document imports for monthly inventory counts, preventing single-batch import timeouts |
Three Common Configuration Errors
- Incomplete data capture, with connection timeout logs displayed:
DB_MAX_LINKis set too low. Concurrent write requests exceed the database connection limit, resulting in partial document data failing to complete writing. - Exported dataset.csv only contains the index field with no content: The storage document content field is not correctly mapped, and core fields such as material code and lot number are not included in index content.
- Slow operation logs are continuously reported during file parsing:
MONGODB_URIis not configured with read-write separation. A large volume of query traffic is concentrated on the primary node, causing response timeouts.
How to Verify Configurations Are Correct
- The database monitoring dashboard is accessed, the cluster node load corresponding to
MONGODB_URIis reviewed, and the read-write request separation ratio is confirmed to meet expectations. - A database connection count query is executed, and the current active connection count is confirmed to not exceed the
DB_MAX_LINKconfiguration limit. - A batch of monthly inventory count documents is imported, and all documents' lot numbers and storage location information are verified to be successfully written to the ledger with no data omissions.
- The dataset.csv for a test batch is exported, and the file is confirmed to include content fields such as material code, lot number, and expiration date.
The values provided are common starting points and should be measured against the reader's own samples.