[cobalt-users] Cobalt bug nameServers???

["Brian A. Chase" <support@xxxxxxxxxxxxx>]

We have seen this from UUnet Cache DNS Servers and several other networks.
We haven't tracked down is it is a communication problem with Cobalt Servers
running bind but it usually clears up in a day.

If you run dig, you will see that the root and cobalt name servers respond
correctly but the UUnet/Worldcom servers do not see the records.  This has
the effect of domains dropping off the net for a day or more.

If anyone has solved this, please post.  Worldcom's response was "Use
another DNS".  This of course is not a valid solution for the other 50
million Worldcom customers.


_________________________
Andy Brown wrote:

> I *think* I've seen this behaviour in some of our Raqs here, but what I
> ended up trying seemed to cure it.
> Out of interest have the machines been rebooted recently, or simply not
> ever had bind/named restarted manually??
> For some reason after a reboot our machines will start the
> /usr/sbin/named process but it won't answer queries in the way that you
> explain.
> If I ssh in and call a
> /etc/rc.d/init.d/named restart
> Give it a minute to settle down, and hey presto it starts to answer
> again, and will continue to do so after that point.

None of what you did should have made a difference to Frderik.

His system isn't using the named daemon on the RaQ at all; the
resolver's going right out the box to uu.net.  The problem is probably,
as he suggests, is time-out based on slow connectivity.

Jeff