incident-response

Determine if a production rollback is needed by collecting current error rates, P50/P95/P99 latency, CPU, and memory usage, then comparing them against baseline thresholds. This process outputs a structured incident assessment with actionable rollback recommendations.

Collect error rates, P50/P95/P99 latency, and CPU and memory resource metrics for on-call triage during an outage. Comparing these against baseline thresholds identifies the root cause and impact to output a structured incident assessment.

Yes, you can assess elevated error rates across deployments by applying threshold analysis to current metrics. Comparing current error rates and latency spikes against baselines determines severity and outputs recommended rollback or mitigation actions.

You should not rely solely on an automated rollback decision when CPU and memory resource metrics remain within normal baselines despite a latency spike. Accurate rollback decisions require comparing all collected metrics against baselines to confirm a deployment is the actual root cause.

Structure an incident assessment for postmortems by outputting a summary that compares current error rates, latency, and resource metrics against baselines. Include threshold analysis results and actionable rollback or mitigation recommendations to guide future prevention.