| Service | 30 days | 90 days | 365 days |
|---|---|---|---|
| REST API | 100.000 % | 99.981 % | 99.964 % |
| Dashboard | 100.000 % | 100.000 % | 99.991 % |
| Webhooks | 99.912 % | 99.887 % | 99.902 % |
| Object storage | 100.000 % | 99.958 % | 99.939 % |
Unclean database failover left stale read replicas in rotation.
A single tenant’s retry loop exhausted the shared queue budget.
A CDN edge in Frankfurt served 502 for static assets after a bad cache purge. Traffic was steered away from the edge; the vendor confirmed the fix on 21 June.
Multipart uploads above 100 MB failed to complete on the first attempt for roughly 6 % of requests. Caused by an undersized connection pool on the new storage cluster.
An expired internal TLS certificate broke service-to-service calls. Certificate rotation is now automated and alerts fire 21 days before expiry.