medium10 pts · first answer countsobservabilityon-call

Choosing what to page on

The on-call engineer gets paged several times a night for "CPU above 80 percent" while users notice nothing. What is the better alerting approach?

A.Lower the CPU threshold to 60 percent and add memory and disk alerts so problems surface earlier.B.Page on every warning and error log line, so that no early sign of trouble is ever missed.C.Page on user-facing symptoms like error rate or latency against the SLO; send CPU to dashboards.D.Remove all alerts and rely on customers reporting issues, since users noticed nothing anyway.

Discussion 0

Join free to post in the discussion.

Choosing what to page on

Discussion 0

Related questions