Over the last two days we have a network problem which is worsening all
the time. It manifests itself fundamentally by a ping or
traceroute failing with
'host unreachable' from the alpha, while at the same time another host on
the ethernet can ping or trace the same address satisfactorily.
The hosts which cause this do not seem to have any pattern eg it can be
any address not all beginning with, say, 206.
We thought it was the dns or the dns cache, but have updated the cache to
the latest, and rebuilt bind with bindsetup without any change. We have
also spent hours at debug level 1 to 3 of bind with responses like resend
multiple times - seeming to indicate that bind is saying I can't get that
nameserver mate, rather than there is something wrong with your
nameserver.
The routing is very simple. There is one gateway - a DECBrouter on
194.72.80.1, and one default route to our backbone on 194.72.10.250. It
is not even all distant networks that fail. EG we can't see our competing
isp on 194.72.192.3 who is 3 hops away, but we can cheerfully ping
gatekeeper.dec.com half the world away.
I have been trying to track this down for over 20 hrs solid and am at a
loss. Can anybody help me out by suggesting avenues to explore please?
We have re-booted three times.
Regards Stuart McKenzie
Received on Fri Mar 28 1997 - 22:33:47 NZST