Recently, the hypervisor was offline (network only) for a few minutes while Telia broke their routing. This was resolved within five minutes. Downtime was well within our SLA.

NS3 was undergoing a complete rebuild on a new node; although we set a low TTL, our monitoring tools cached it for nearly 20 minutes; this affected nothing but monitors, as the old node was powered off on July 12th, 2019- nearly a week after decomission.