Observability · On-call judgement
The pager everyone learned to ignore
Alert volume doubled and on-call is ignoring pages. How do you fix it?
Monday 11:00 · on-call retrospectiveIllustrative scenario, not a real incident
A real outage waited 40 minutes for an acknowledgement, on a pager that fires seven times a night
Pages to the platform rotation have doubled in three months. Most of them resolve by themselves, and engineers have learned to silence the phone and go back to sleep.
Last Thursday one of those pages was a real outage. It looked like the others, and nobody opened it for 40 minutes.
What the dashboards and logs show
- Pages last month
- 212, up from 104 a quarter ago
- Median time to acknowledge
- 19 minutes, up from 4
- Last Thursday's outage
- opened 40 minutes after the page
Clarify · optional · scored under "Hypothesis before change"
Ask the interviewer
0 of 3 asked
Each question reveals a fact. What you ask before you act is part of your score.
First move · scored under "Mitigate before root cause"
Tonight's shift starts in six hours, with the same pager. What do you do first?
Your rubric score, the model answer, the follow-up questions and the related lessons appear once you have answered the last step.