Railway GCP Account Suspension Incident Report

13 hours ago

Railway experienced a platform-wide outage lasting approximately 8 hours on May 19-20, 2026, due to Google Cloud incorrectly suspending their production account.
The suspension immediately disrupted GCP-hosted infrastructure, including the dashboard, API, control plane, and databases, causing 503 errors and login failures.
As cached network routes expired, the outage cascaded to workloads on Railway Metal and AWS, leading to widespread 404 errors and rendering all regions unreachable.
Recovery involved restoring GCP account access, persistent disks, compute instances, and networking, with services gradually restored over several hours.
During recovery, GitHub rate-limited Railway's OAuth and webhook integrations due to a surge in retry requests, temporarily blocking logins and builds.
Railway takes responsibility for architectural dependencies that allowed a single provider action to cause a full-platform outage and is implementing changes to prevent recurrence.
Planned improvements include removing the dependency on GCP for the network control plane, extending high-availability database shards across AWS and Metal, and removing Google Cloud from the data plane's hot path.

Hasty Briefsbeta