Reconcile and drift statuses
Background on both is in keeping stock in step.
Reconcile reviews
Written when a reconcile finds a difference between the store's on-hand figure and Yessort's total for that site.
| State | Meaning |
|---|---|
| Open | Awaiting a decision, or a record of an automatic booking you have not acknowledged |
| Resolved | Dealt with, with who resolved it and when |
| Dismissed | Closed without action |
Each row records the store's figure, Yessort's figure, the signed difference between them, and whether it was booked.
| Booked | Meaning |
|---|---|
| Yes | The difference was written to stock automatically. The row is the visible record of it |
| No | Nothing was written. The difference is held for you |
A row is not always a job. Some are a record of what Yessort did, kept visible on purpose.
Repeating a reconcile for the same variant, site and cause refreshes the existing row rather than adding another — re-opening it if it had been resolved.
Which reconcile booked it
| Reconcile | Behaviour |
|---|---|
| Channel activated with listings already linked | Books nothing. Every difference is held |
| Review queue decision applied | Books a difference within 20% of the store's figure, or five units, whichever is larger. Holds anything larger |
Drift passes
Compare what the store is showing against what Yessort holds. They report; they change nothing.
| Mode | When | Coverage |
|---|---|---|
| Sample | Nightly | A bounded subset of the channel's live listings. Cheap, so it never contends with other work |
| Full | Weekly | Every live listing, from one bulk export |
| Status | Meaning |
|---|---|
| Completed | The pass ran |
| Skipped | Deliberately not run — usually because the store already had a bulk export in flight. Not a failure |
| Failed | Stopped, with a reason |
A store runs one bulk export at a time, so a full pass can find the slot taken by a catalogue sync. Treating that as an error would raise an alarm every time the two overlapped.
Each pass records how many variants were checked, how many variant-and-location pairs were compared, and how many diverged beyond the threshold. Where a pass covered less than everything — a skipped slot, or a cap on how many variants it looked at — that limitation is recorded alongside the result, so a partial sweep is never mistaken for a clean bill of health.
Every pass is recorded whether it found anything or not. That makes "this channel has not been checked since Tuesday" answerable — a check that quietly stopped running looks identical to one finding nothing, unless the runs themselves are kept.