The incident was resolved after 36 minutes (Oct 18, 2021, 17:57 UTC). The incident was caused by administrative (human) error.
As a result, it was necessary to roll back our databases to the state as of Oct 18, 2021, 17:00 UTC. API write operations between that time and 17:21 UTC are lost and unfortunately cannot be recovered. We apologize for any inconvenience.
DNS operations continued throughout (that is, domains hosted at deSEC experienced no outage).