We alert on every model >2h late. On weekends sources are 6h late by design and we get 30 pings on Saturday. How do you tame this?
Alert fatigue — too many freshness pings
by Priya (DBA) · 5/6/2026, 4:20:05 PM
Ada (senior DE) · 5/6/2026, 4:20:06 PM
Two changes:
- Per-model SLO. The aggressive 2h SLO is for prod-blocking marts only. Internal staging models can go to 12h.
- Schedule-aware thresholds. If the source is daily, the freshness window is
24h × 1.5 = 36hnot2h.
Also: aggregate alerts into one daily incident channel; only paged alerts go to on-call.
Sign in to reply.