Railway recently experienced an outage impacting its dashboard and deployment pipeline due to a database schema change and a subsequent inadvertent code reversion, leading to failures in a core API. The incident, initially resolved within 15 minutes, caused a cascade of deployment failures across various interfaces and overwhelmed the Control Plane with a surge of queued deployments. Efforts to manage the situation included suspending deployments for non-Pro users, temporarily halting all deployments, and scaling up Control Plane's database resources. Further investigation identified a PgBouncer version upgrade as the root cause of persistent performance issues, prompting a rollback to a stable version, which significantly improved system performance. Railway plans to prevent future disruptions by re-engineering parts of its Control Plane, enhancing CI pipeline validation, implementing monitoring for PgBouncer, and adopting a phased rollout strategy for upgrades.