The dns server mask 18.104.22.168 which we propose to our dedicated
servers' customers is a distributed system which independently manages
every /24. The aim is to avoid the overloads.
Since the maintenance tasks on the MRTG servers (the kernel updates),
we have noticed that sometimes these DNS servers go down in a totally
independent manner and provoke a mask failure for a specific /24.
All of the rest continue to function.
We are introducing a check system and a reset of CDNS
in case of problem or/and routing change in order to access
to cdns which functions (in case of problem on another).
Date: 2010-06-08 00:18:52 UTC The problem is solved.