Predictable traffic spike causes 503s

A ticketing service has a known traffic pattern: every weekday at exactly 12:00 a popular daily on-sale opens and traffic spikes 5x within 60 seconds. Every day at 12:00 you get a 3-4 minute window of elevated p99 and some 503s, then it settles. Dashboards show CPU saturating right at 12:00, the HPA scaling up but new pods only becoming ready ~2-3 minutes later (image pull + app warmup), and the load balancer 503ing while pods are unready. No bug, no bad deploy, it's the same every day. Triage and propose durable mitigations.