Severity
MEDIUM-HIGH: CPU at 99% with sharp upward trend; memory at 95%; immediate action needed to prevent cascade failure.
Root Cause
- Runaway process or uncontrolled workload spike consuming CPU/memory
- Resource leak or inefficient query generating sustained load
Actions
- SSH to affected host; run
top -b -n 1 | head -20 to identify CPU/memory hog process.
- Kill or throttle offending process; if test/batch job, pause execution.
- Check recent deployments (last 2h); rollback if correlation exists.
- Drain traffic to other instances if load-balanced; restart affected service if isolated.
- Increase memory/disk allocation if sustainable load, or tune thresholds if false alarm.
Watch
- CPU trend direction (target: <70% within 5min; confirm downward vector)
- Memory pressure and disk I/O (watch for secondary spikes indicating queue buildup)
Escalate If
CPU does not drop below 80% within 10 minutes, or memory continues climbing.