Between 17:00 – 17:30 UTC an internal data-backfill job briefly saturated our primary database connections, causing intermittent errors and slow loads in embeds and the dashboard. The job was moved to a replica and capacity freed; all services are now fully operational.
[Timeline] • 17:10 UTC — Users reported slower loads or occasional errors in embeds and dashboard pages. Engineers are investigating. • 17:35 UTC — Root cause identified — an internal data-backfill task briefly saturated our primary database’s connection pool, slowing requests. We are throttling that job and adding capacity. • 17:50 UTC — Mitigation applied: we moved the backfill job to a read-only replica and freed up primary-database connections. Errors and slow loads are clearing. • 18:25 UTC — Traffic has been stable for 15 minutes and error rates are back to baseline. A brief summary of the incident will follow.
[Impact] A subset of API, embed, and dashboard requests saw elevated latency while the primary database connection pool was saturated; read-only traffic and most background jobs remained within normal thresholds.