- Shipped
- September 23, 2026 at 1:51 AM UTC
- Author
- Kamo
- Commit
- 77ee843
The Lead Intake settings list showed total_received / total_processed / last_received_at straight off the endpoint row, where they were bumped once per payload on one shared row, lost updates to YugabyteDB 40001 aborts (the bumps' failures were swallowed on purpose) and were clobbered by every settings save — "imported" was 437 short on two endpoints. - create_lead_intake_ledger.sql + **************** a total per endpoint plus an append-only journal written by a trigger on lead_intake_raw_payloads in the payload's own transaction. Applied and backfilled 2026-09-22 (32 endpoints, 0 mismatches). - LeadIntakeController: every endpoint response (list, create, update, toggle) passes through **************** — one ledger read for the whole list. - LeadIntakeReceiveController / processBacklog: the endpoint-row bumps are gone; the payload insert and its IMPORTED status are the counts. An accepted payload can no longer race a stats write at all. Needs kamo-shared-library 78a3f605.