grafclouds.com / documents / training / devops-challenges / noisy-alert-fatigue

On-Call Is Drowning in Alerts. Silence Them?

Intermediate ~12 min read
Observability

Scenario

A legacy alert pages on-call whenever CPU crosses 80% for a minute. It fires several times most nights. Every time, latency and error rates are normal, users notice nothing, and the responder acknowledges it and goes back to sleep — a little more tired and a little more trained to assume the pager is crying wolf. Nobody remembers who created the rule or what it was supposed to catch. The team is worn down, and the on-call rotation is becoming something people dread.

The Quick Fix on the Table

A justifiably exhausted teammate wants to slap a permanent silence on the alert — or just delete the notification route. Thirty seconds of work, and tonight everyone sleeps.

Interview · Round 1

The quick fix is on the table and the room is waiting for your call. Would you sign off on it? Take a position and justify it — out loud or on paper — before revealing the analysis.