Question

Your catalog service calls an upstream pricing service over HTTP. Starting 13:20, ~12% of product pages render with prices missing or showing $0.00, but your error rate is FLAT — no exceptions, no timeouts. Dashboards: the upstream returns HTTP 200 on the affected calls, but the response bodies are truncated or empty (your JSON parser silently yields a default-zero price rather than throwing). The upstream sits behind a load balancer that was reconfigured at 13:15 to add response buffering with a new max-body / idle-timeout setting. How do you triage and mitigate?