Advisories are published to @privatebyte and on the security page
Incidents that affected customer services, with the cause and what changed afterwards.
Inbound traffic lost to a group of addresses in AmsterdamA virtual switch restart cleared the flow table fourteen hours earlier without causing an immediate fault. A later routine operation reapplied a partial rule set, interrupting inbound delivery. Flow state is now restored automatically on service start and reconciled every five minutes, with an alert on drift.
ResolvedOne IPv4 prefix unreachable for approximately 30 hoursStale routing registry objects left by a previous address lessor took effect when a transit provider rebuilt its prefix filters. Outbound traffic was unaffected; inbound was not delivered. The objects were removed and filters rebuilt. Registry and RPKI records are now checked daily.
ResolvedPortal account-identifier confusionA mapping error meant a small number of accounts could briefly see another customer’s services within a session. Patched the same day and reported to the ICO within the statutory window. A regression test is pinned in CI.
Resolved