Fly.io Outage History

Fly.io experienced networking issues in their Sydney (SYD) region that affected multiple central services for approximately 24 hours. The engineering team identified the root cause and applied a fix to restore normal operations. All services were confirmed to be working normally after the resolution was implemented.

Fly.io experienced heightened network latency in their ORD (Chicago) region for 4.3 hours. The latency issues affected network performance for services running in that region. The incident was resolved after implementing a fix and monitoring the results.

Fly.io's Managed Postgres control plane in the NRT (Tokyo) region experienced instability for 3.1 hours, causing unexpected cluster failovers that resulted in dropped connections and degraded performance for users with clusters in that region. While clusters automatically returned to health after each failover, users experienced intermittent service disruptions during this period. The issue was identified and resolved with a fix that normalized MPG performance in the affected region.

Some hosts in Fly.io's Chicago (ORD) region became inaccessible, also affecting a small number of Managed Postgres clusters. The incident was resolved after 48 minutes of working with their infrastructure provider.

Fly.io experienced control plane issues affecting Managed Postgres clusters in the SYD region for 1.7 hours. While most clusters continued running normally, new database creation, backup restores, and upgrades encountered errors or delays, and some clusters experienced failover events from primary to standby nodes. The incident was resolved after control plane operations returned to normal and all clusters were restored to healthy status.

Fly.io experienced a major incident with their metrics cluster that lasted 3.9 hours. The metrics system was affected, preventing users from accessing performance and monitoring data. The issue was resolved after implementing a fix and monitoring the cluster for stability.

Fly.io experienced a major outage affecting their GraphQL API and web dashboard for 3.2 hours. The incident impacted dashboard functionality and caused delays in app state updates even after initial restoration. The issue was resolved through an implemented fix that restored full service availability.