Date: March 7, 2026
Duration: ~25 minutes (13:45–14:10 CET)
Impact: ~30% of funnels experienced slower navigation between screen, while syncing with the Funnel API
On March 7 at 13:45 CET, about 30% of funnel pages were loading slower.
The issue lasted approximately 25 minutes and was fully resolved by 14:10 CET. Our CDN cache continued serving most funnels normally, which reduced the visible impact.
A spike in internal database queries on our read replica created replication lag between the replica and the primary database.
While lagging, the replica served outdated data for some funnels — and those stale responses got cached by our CDN, extending the delay beyond the lag itself.
We switched funnel traffic from the lagging replica to the primary database, which restored availability immediately. We then purged the cached empty responses. We will conduct an internal investigation how to avoid such incidents in the future.