Question

Your team runs a payment API on GKE across three regions. Users in one region report intermittent failures, but your primary availability alert (aggregated across all regions) has not fired because the global success rate remains above the SLO threshold. You need a dashboard that lets on-call engineers quickly identify which region is degraded without altering your existing global SLO alert. What is the most effective approach?

Accepted Answer

Create a dashboard chart of the request success-rate metric with an aggregation that groups the time series by the region label, displayed as separate lines per region.

Answer

Lower the global availability SLO target so the aggregated alert fires whenever any single region degrades.

Answer

Add a metric-absence condition to the existing global alert policy so it triggers when regional data stops arriving.

Answer

Replace the request success-rate metric with a log-based counter metric that counts only 500 errors globally.