Email sub-processor returned 503 to ~14% of outbound traffic. Failover engaged in 22m. No data loss; affected sends auto-retried within 1h. Root cause: upstream provider regional outage. Mitigation: dual-provider failover now ships with default config.
Brief pause on EU-West signal surfacing during a planned read-replica failover. Stream caught up within 11m, no missed signals (held in buffer).
The pipeline cockpit took 4-6s to first paint for ~9% of users. Caused by a CDN cache invalidation cascade. Fixed; added regional pre-warming.
HubSpot OAuth token rotation broke for accounts with custom scopes. Forced reauth notice sent to 31 affected workspaces. No third-party data exposed.