During updates of lower campus VM host cluster,
the experimental Kea DHCP nodes,
which are currently providing DHCP for house N in a field test went down.
Normally, the HA mechanism should have prevented this,
as one node was kept alive at all times.
However, the first node didn't start again
due to a software bug killing the DHCP server if it was unable to connect to the local database
and the system not having automatic restarts on failure enabled.
As soon as the problem was noticed, it could be fixed by simply restarting the nodes,
as they now were able to connect to their database servers.
