A reboot of one of our Hubject services caused a couple of minutes downtime.