Damaged Returns Location. Where returned stock graded non-restockable by the warehouse is received. Leave empty and the connector creates a Returns - Damaged internal location under the mapped warehouse the first time it is needed, then remembers it here. Point it at your own location if you already operate one. Restockable lines always go to normal sellable stock. See Returns for how grading works.
Auto-validate GRNs (default on; shown with GRN support). On: when the 3PL confirms goods received, the Odoo receipt is reconciled to the reported quantities/batches and validated immediately; shortages become a backorder receipt so nothing is silently dropped. Off: the reported detail is posted to the receipt's chatter and an activity asks a person to review and validate - the right choice while you are still building trust in a new 3PL's counts.
Stock Master (default 3PL is Master). The single most important inventory decision. 3PL is Master: inventory syncs set Odoo's stock to match the warehouse count, using native, fully-auditable inventory adjustments, batch by batch. Odoo is Master: syncs never change Odoo stock - every difference is recorded as a discrepancy line on the sync report for a person to resolve. Example: most 3PL businesses run 3PL-master (the goods physically live there). Run Odoo-master temporarily during a stock dispute: you keep full visibility of what the 3PL claims without letting it move your books.
Stock Adjustments (default Apply Immediately). Only meaningful with 3PL-as-master. Apply Immediately: differences become posted inventory adjustments during the sync. Leave as Counted Quantities for Review: differences are staged as counted quantities in Inventory > Physical Inventory where a stock controller applies or rejects them - a half-way house that keeps human sign-off while the 3PL still supplies the numbers.
Run Flows as Queue Jobs (default off).
With the OCA queue_job module installed on a Linux server, each
operation (each order push, each status check…) runs as an individual,
deduplicated background job instead of inline in the scheduler -
better throughput and isolation for high volumes. The switch explains
itself honestly: it is inert on Windows hosts (the OCA job runner is
POSIX-only) and hidden when queue_job is not installed. Failures
still land in the Sync Log with one-click retry either way.
Keep Logs (days) (default 90). Sync Log entries older than this are deleted nightly. Set 0 to keep everything forever. Chatter history on instances, orders and returns is never touched by this - only the technical API log.