Our old healtcheck system was prone to false positives during high server load events which caused false positives on the uptime monitor.
We've updated the health check to be more resistant to high load events and have reset the uptime statistics to better reflect true uptime statistics.