July 2026 Summaries

Feature flags have been introduced on Railway to provide a safer and more flexible way to manage feature rollouts in production environments. This approach allows developers to toggle features on or off at runtime without requiring redeployment, facilitating a controlled and gradual release process. By integrating feature flags into the dashboard, CLI, SDK, and MCP, developers can create flags, set default values, and define targeting rules to specify conditions under which features are enabled for specific user groups. This system simplifies the process of rolling out new features incrementally, starting with internal teams and expanding to a percentage of users, while allowing for quick rollbacks if issues arise. Feature flags act like dynamic environment variables, updating without redeploys and offering clear, reproducible evaluations of why a user sees a particular feature state. The system prioritizes default values to prevent unexpected behavior from conflicting rules, ensuring a stable production environment. While currently resolving to literal values, Railway is exploring further applications of this context-based resolution system.

On July 2, 2026, Railway experienced a significant outage in one of its US East availability zones due to a network degradation in an ISP, leading to increased latency and packet loss between US regions. Efforts to reroute traffic inadvertently left the zone without a stable internet route, exposing hidden bugs and causing degraded disk performance and private networking issues for approximately two hours. The incident unfolded as a combination of ISP degradation, mismanaged routing changes, and system behavior flaws, which resulted in traffic being diverted onto slower backup networks. Despite the initial response to reroute traffic, issues persisted due to lingering connections on incorrect paths and misconfigured private networking tunnels. Corrective actions included terminating stuck connections, restarting mesh networking agents, and ensuring future resiliency through infrastructure updates and improved alert systems. The incident highlighted the need for updated designs in older sites and the importance of proactive routing management to prevent similar disruptions in the future.