Railway incident RL8FRJE6 matches an active production delivery failure for Verotta Insight V2.

Project: d2ce415e-1b9d-44f1-adc2-1846eee171da

Service: Verotta Insight V2 / web / production (attached)

Domain/region: app.verotta.com / US West

Credits are available: the Trial workspace shows 25 days or $4.47 remaining, $0.53 used, and a $0 current bill. All four services are online. The health endpoint returns exact application SHA 11dce8b9ead2f84d47c084a4694511c0909ccc25 with PostgreSQL, Redis, and worker all ok. Web CPU remained near idle, memory was approximately 120-150 MB, and there was no container crash, restart, or application error. Application code and configuration were unchanged.

Railway Network Logs for deployment c7529982-e2b8-48ef-a0ba-d9769280be59 recorded 26 proxy-generated 502s, nearly all at the 15-second timeout boundary, across /, /api/health, application pages/APIs, favicon.ico, and the exact Next.js CSS asset. The CSS failure produced an unstyled page, and a 502 HTML response to an API poll caused a JSON parse error.

We redeployed the exact same SHA and configuration as deployment 36c9d9ad-7473-4524-8d5f-af6f07b75f32. It became Active after migrations and a clean Next.js start. After 24 uncached health/root/CSS requests returned 200, a health request returned 502 after 3.5 seconds; the immediate retry returned 200 with the same exact SHA and all dependencies ok.

Please investigate the US West public edge-to-replica routing/timeouts for this service and rehome or migrate it from the affected host/network path if needed. Please confirm when the path is stable.