Alert: HighCPU

Symptom

Node CPU above 85% for 30 minutes.

Immediate action

Identify instance from alert labels:

ssh <host> 'top -bn1 | head -20'

Diagnosis

1. Expected load on inference-01 during model inference.
2. Backup or batch jobs.
3. Runaway process.

Escalation

Ryan if sustained on lastvet-api, lastvet-db, or lastvet-edge.

Resolution history

Date	Notes