"Why didn't I get notified?" always has an answer in the product. Work the ladder:
- Recent deliveries. A
deadorpendingrow with an error means the receiver refused or was unreachable — the stored error is the receiver's own answer, secrets redacted. Fix, then Re-drive. - A
suppressedrow — noise control worked as configured; the row names the reason. Evidence, not silence. - No row at all — either no rule matched (check types and minimum severity against the catalog) or the event never fired (check the audit chain). The classic cause is a drift line: a rule naming a channel document that does not exist — legal, because apply order never matters, and flagged until the channel arrives.
- Queue stuck — pending count growing, oldest age climbing: is the notifier container running? Its log has an edge-triggered "delivery failing" line naming the cause.
- The same alert twice — by design, rarely: delivery is at-least-once. Receivers that must not repeat dedup on
X-GOCA-Delivery.
Secret-reference errors have their own rows — a missing stored secret, a wrong recipient, a regenerated recipient key — and each names the re-entry fix.
Now I can answer "why didn't it page" from the delivery history.