| Date | Region | Impact | Duration | Summary |
|---|
| 2026-10-29 | fra1 | Degraded writes | 27 min | Metadata service leader election took longer than usual; a subset of writes returned 503. |
| 2026-10-16 | hel1 | Read errors | 9 min | Cache node restart after a kernel update; reads failed over to ams1 automatically. |
| 2026-10-03 | fra1 | Partial | 18 min | Multipart complete returned 503 during a rollback; retries succeeded after the deploy settled. |
| 2026-09-20 | ams1 | Elevated latency | 42 min | Upstream switch maintenance caused retries on the write path; clients saw p99 above 400 ms. |
| 2026-09-07 | hel1 | Partial | 16 min | Replication lag exceeded the alert threshold after a large customer upload finished. |
| 2026-08-25 | ams1 | Elevated latency | 25 min | A backup job saturated the metadata disk; read latency rose until the job was rescheduled. |
| 2026-08-12 | fra1 | Partial | 11 min | A bad gateway deploy was rolled back; the error rate returned to baseline. |