We are unable to bring one of our linux boxes into the network once it was restarted. But once we stop and start the network service. We are able to ping the gateway and get into network. After a minute or two we get the following reponse for the ping command

64 bytes from 124.168.215.40: icmp_seq=0 ttl=116 time=206.8 ms (DUP!)
64 bytes from 124.168.215.40: icmp_seq=0 ttl=115 time=216.8 ms (DUP!) ...

After this the server gets disonnected from the network.

Please help in resolving this issue.

Regards,
Antony

hello,

you have duplicate IP inside your network.

cheers

Thanks for the reply.

Even some of my friends suggested that there is a duplicate IP.

The link in the ethernet card was in Half duplex mode. This was the issue. Once it was changed to Full Duplex the issue was resolved. But do u have any idea why did it give a message which is similar to the duplicate IP?

hello,

the theory of CSMA/CD - NIC will listen to link whether it is clear to transmit data or not. If it founds any traffic it will wait and check for whom the traffic destined - otherwise its turn to send data. If at the time it send data and anyone else do the same - it will drop the traffic and tell you DUP.

BUT - this is not the case - since modern NIC or OS have autonego option (or even auto-MDIX for switches) - it listen to the HUB/SWITCH what mode they operate.

i think you still should check any DUP IP.

cheers...

Hi,

The issue is not there anymore...and I am sure there is no duplicate IP in the network. Also I cross checked the same with our network team.

Regards,
Anony

We have a similar problem in some linux boxes on our network, they have dual ethernet with vlan support through bonding driver in active-backup mode on a cisco catalyst switched LAN (with trunk support). Although we think that this is some problem with the multicast filter on the switch, we can't still find out what the real problem is.
The other case is a strange delay after starting both interfaces, that takes about 30 seconds to being able to ping any host on the LAN... I've heard that is a "normal behaviour" on some Cisco switches.