ModernLoop Slowness

This incident has been resolved.

Monitoring — Slow loading of task pages, the dashboard, and scheduling, and intermittent timeouts, have been resolved.

Starting at 10:15 UTC (3:15 AM PT), a subset of customers experienced slow page loads and occasional timeouts when opening task pages, the Upcoming Events dashboard, Reporting, and the scheduling flow. The public API and MCP integrations could also return timeout errors during this period. No data was lost, and no scheduled interviews were affected.

We identified the cause: a database read path that, for some accounts, repeated the same large query many times in a single request and kept running after the request had already timed out. This placed sustained load on our read database replicas. We deployed a fix at 15:20 UTC (8:20 AM PT) that limits that path, and response times returned to normal by 16:00 UTC (9:00 AM PT).

We are monitoring the platform closely while the remaining database load returns to normal, and we are rolling out additional safeguards so that a single request can no longer create this kind of load. We will post a final update once monitoring is complete.

We apologize for the disruption.

We are continuing to investigate this issue.

We are currently investigating this issue.