Find the answer to your Linux question:
Results 1 to 2 of 2
I'm hoping someone here can help me understand this problem: I have a server with 4 Gb Ethernet interfaces. All interfaces are connected via a 24-port switch to other servers ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
  1. #1
    Just Joined!
    Join Date
    Jul 2007
    Posts
    2

    targetting specific NIC on multi-homed host


    I'm hoping someone here can help me understand this problem:

    I have a server with 4 Gb Ethernet interfaces. All interfaces are connected via a 24-port switch to other servers in the lab and the uplink. Each interface has a statically assigned IP address and appears to be active.

    When SSH'ing to any of the four IP addresses on this box, it appears that all SSH traffic is going through eth0. tcpdump -i eth0 displays the SSH packets on eth0, even though the SSH session is directed to the IP address of eth1, eth2 or eth3.

    Similarly, when pinging the IP addresses of eth1, eth2, or eth3, tcpdump shows that all the ICMP packets are being handled by eth0 rather than the interface associated with the IP address.

    What I'm expecting is that pinging the IP address of eth1 will result in ICMP packets incoming on eth1, pinging eth2 will result in ICMP packets on eth2, etc. I'm at a loss as to why packets directed to the IPs of eth1,2 & 3 are all being handled by eth0.

    My next step will be to examine the ARP tables of the switch, although I'm not quite sure what to look for. I don't have a lot of experience with network configuration. Any help / explanation would be appreciated.

    Again, I'm interested in targeting a specific interface on a server with multiple GbE connections. I would like to direct a stream of TCP/IP client data directly to a specific interface on the server by specifying that interface's IP address in the client program.

    Thanks!

  2. #2
    Just Joined!
    Join Date
    Jul 2007
    Posts
    2

    found the answer...

    Hi everyone, thanks for reading my question but it turns out that the problem was caused by ARP flux. Basically, the ARP responses were being sent out on the first available interface, leading to all sorts of weirdness in the ARP caches of machines on the subnet.

    This page provided a solution:

    https://tier2.ucsd.edu/t2/index.php?...id=27&Itemid=7

    Cheers

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •