Wangdiantong Other Inbound Orders Sync to Kingdee Cloud Galaxy: Field-Proven Configuration and Pitfall Review for Supply Chain Integration
What This Strategy Solves
In one retail customer's on-premise setup, hundreds of other inbound orders are generated daily in Wangdiantong·Qijiqimen, while the financial books and inventory ledger live in Kingdee Cloud Galaxy. When the two sides drift apart, month-end closing drags into the early morning. On the customer's site we used the Qeasy Data Integration Platform to get this single document-type pipeline running cleanly, so other inbound orders flow from operations straight into finance without manual re-entry. This article breaks that one strategy down end to end.
Data Flow and Field Mapping
The overall direction is Wangdiantong·Qijiqimen → Qeasy middle layer → Kingdee Cloud Galaxy. The middle layer is not a passive pipe — it owns encoding conversion, field trimming, and exception parking.
Key field mapping table:
| Business Meaning | Wangdiantong·Qijiqimen | Qeasy Middle Layer | Kingdee Cloud Galaxy |
|---|---|---|---|
| Document No. | order_no | doc_no (unified prefix) | FBillNo |
| Document Type | type_code | doc_type (dict mapping) | FBillTypeID |
| Warehouse Code | warehouse_id | wh_code (unified) | FStockID |
| Owner Code | owner_id | owner_code | FOwnerID |
| Supplier Code | supplier_id | sup_code (centralized) | FSupplierID |
| Material Code | sku_code | mat_code (unified) | FMaterialID |
| Quantity | qty | qty | FQty |
| Price | price | price | FPrice |
| Note | remark | remark | FNote |
Encoding mapping must be centrally managed in the middle layer. We keep separate mapping tables for warehouse, supplier, and material codes under Qeasy's Encoding Center. The strategy only references encoding variables — no hard-coded strings live in scripts.
How to Configure in Qeasy
After signing in to the Qeasy console, follow these steps to wire up the pipeline:
- Create an integration pipeline: pick the source adapter "Wangdiantong·Qijiqimen" and the target adapter "Kingdee Cloud Galaxy".
- Capture strategy: enable "by document status, incremental + full snapshot". Other inbound orders are only pushed after posting, so the source uses the "approved" timestamp as the incremental anchor.
- Middle-layer processing: add a "Code Mapping" step inside Qeasy's data processing node, reference the Encoding Center tables, and enable "header/body staged push" — push the header first, retrieve FBillNo, then write the body back. This avoids orphan body rows with no parent document.
- Write to Kingdee: on the Kingdee side, target the "Other Inbound Order" document type and call its open API. On failure, Qeasy retries 3 times at 5-second intervals; persistent failures go to an exception queue for manual handling.
- Logging and reconciliation: turn on "per-document reconciliation". Qeasy produces a daily diff report showing which documents exist on the source side but not the target.
Implementation Steps
We recommend rolling this out in three phases:
- Incremental kickoff: start with the last 7 days of approved orders in a small batch to validate field mapping, code conversion, and document type before opening the floodgate.
- Full backfill: shift the starting point to the business system go-live date and replay all historical other inbound orders. In one customer run this took about 4 hours for 12,000 documents. Kingdee's concurrency limit was hit once; Qeasy auto-throttled to 30 documents per minute without human intervention.
- Scheduling frequency: set the daily schedule to every 15 minutes with a 30-minute incremental window so business peaks don't pile up. The reconciliation job runs daily at 1:00 AM.
This is the "incremental + full dual-track" playbook we've validated repeatedly: full backfill during cutover, then incremental only once steady-state.
Pitfall Review
- Body written before the header: a classic mistake is submitting header and body together, which Kingdee rejects because FBillNo isn't generated yet. The safe pattern is header-first, retrieve the document number, then write the body back.
- Encoding mapping scattered in scripts: every environment migration becomes a rewrite. Move all mappings into Qeasy's Encoding Center so strategies reference variable names and migration cost approaches zero.
- Ignoring Kingdee concurrency limits: Kingdee Cloud Galaxy on-premise has a low default write concurrency; pushing the full backlog at full speed will get you kicked. Enable Qeasy's adaptive throttling node — it backs off automatically on 429 responses.
- Wrong incremental anchor: we once saw teams use "creation time" as the anchor and miss the midnight batch. Switching to "approval time" fixed it.
- No exception parking configured: if a document is rejected for a field-length overflow and there is no exception queue, that record disappears. Configure "failed → queue + email alert" in Qeasy so ops can clear it the same day.
Suitable and Unsuitable Scenarios
This strategy fits on-premise deployments with multiple coexisting business systems where front-end operational documents need to sync to the finance or ERP system in near real time. It is not suited to complex transfers that require human decision-making in the loop, nor to scenarios demanding sub-second latency — minute-level sync is already more than enough for other inbound orders.