A caching regression on our edge layer caused public status pages to serve stale data for up to 90 seconds. Dashboard and alerting were not affected. Resolved by rolling back the edge cache config.

The Amsterdam check region missed two scheduled probe cycles due to a scheduler restart. No alerts were dropped; checks resumed automatically.

Slack's incoming webhook API rate-limited a batch of alert deliveries during a traffic spike. Webhook and PagerDuty delivery were unaffected. We've since added exponential backoff for this integration.