WHAT'S GOIN' ON HERE!!! (a little lion king reference...)


OK, here's my dilema. I have 2 servers:

server1:
Xen Linux on kernel 2.4.30
eth0 = 192.168.1.2
eth1 = 192.168.4.4
NIS Master for 192.168.4.0 network
no default route, using policy based route maps

server2:
Red Hat ES 3 on kernel 2.4.21-32
eth0 = 192.168.0.2
eth1 = 192.168.4.2
NIS Client on 192.168.4.0 network
no default route, using policy based route maps

NFS mounts and logins worked perfectly for 3 weeks, then server2 started to hang on login. I tracked the problem to the fact that server2 can no longer see server1 on the network. After turning off ypbind I can log into server2 normally. Both servers are connected to the same HUB (not switch).

From server1 I can ping everything on 192.168.4.0 except server2.

From server2 I can ping everything on 192.168.4.0 except server1.

Both servers accept incomming ssh connections on their 192.168.4.x interfaces.

I have no iptables configured.

Both servers are Dell PowerEdge 2850 servers.

Anybody know why a linux host would stop seeing a specific computer on the network when it can see all other computers/devices on the same subnet?

faulty hub port? I've seen it before when one port on the hub/switch fails yet all the others work perfectly.

I think I would do a traceroute first to see exaclty how things are being routed before hypothesizing a faulty port.

from both servers I can ping all other devices on the hub, so the hub port should be OK.

And a traceroute won't work because I'm on the same subnet.