Reported downtimes were not actual downtimes. The reported nodes where misconfigured with BetterStack for a few seconds at the time of the incident (IP rules).