Results 1 to 10 of 10

Thread: Weird multi-home issue....

  1. #1
    Join Date
    Apr 2006
    Beans
    41

    Weird multi-home issue....

    I have two physical nics in a NAS box I built.... I have a switch that has been configured with two VLANs and this is working on my ESXi servers. But my NAS box will only communicate on one network and the config looks correct.... See below:

    root@pigpen:/export# netstat -anr
    Kernel IP routing table
    Destination Gateway Genmask Flags MSS Window irtt Iface
    0.0.0.0 192.168.0.1 0.0.0.0 UG 0 0 0 eth0
    192.168.0.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
    192.168.100.0 0.0.0.0 255.255.255.0 U 0 0 0 eth1
    root@pigpen:/export# ifconfig
    eth0 Link encap:Ethernet HWaddr 00:26:2d:03:0c:7e
    inet addr:192.168.0.111 Bcast:192.168.0.255 Mask:255.255.255.0
    inet6 addr: fe80::226:2dff:fe03:c7e/64 Scope:Link
    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
    RX packets:72494422 errors:0 dropped:0 overruns:0 frame:39
    TX packets:6671445 errors:0 dropped:0 overruns:0 carrier:0
    collisions:0 txqueuelen:1000
    RX bytes:106679526386 (106.6 GB) TX bytes:553000991 (553.0 MB)
    Interrupt:16

    eth1 Link encap:Ethernet HWaddr 00:1b:21:a0:b9:f5
    inet addr:192.168.100.20 Bcast:192.168.100.255 Mask:255.255.255.0
    inet6 addr: fe80::21b:21ff:fea0:b9f5/64 Scope:Link
    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
    RX packets:1522 errors:0 dropped:0 overruns:0 frame:0
    TX packets:56 errors:0 dropped:0 overruns:0 carrier:0
    collisions:0 txqueuelen:1000
    RX bytes:114833 (114.8 KB) TX bytes:5765 (5.7 KB)
    Interrupt:16 Memory:f0320000-f0340000

    As you can see, I have no gw set up on the 192.168.100.xxx network (eth1), which is the private network for storage in my virtual environment. I cannot ping on eth1 anywhere on 192.168.100.xxx but eth0 works perfectly.

    OK folks, I am not new to Linux (not even remotely) but WTH? LOL

    EDIT to add: I am using ports that are properly configured for the appropriate VLAN.
    Last edited by tonyric; May 13th, 2012 at 11:52 AM.

  2. #2
    Join Date
    Aug 2010
    Location
    hole
    Beans
    415
    Distro
    Ubuntu

    Re: Weird multi-home issue....

    you can start with a tcpdump when pinging from another machine and if there is no traffic dumped can be a vlan issue.
    if it indeed dumped you can get into the higher application levels such as firewall or others.

  3. #3
    Join Date
    Nov 2006
    Location
    Belgium
    Beans
    3,023
    Distro
    Ubuntu 10.04 Lucid Lynx

    Re: Weird multi-home issue....

    what elica says.

    or
    EDIT to add: I am using ports that are properly configured for the appropriate VLAN.
    Are you sure ? Did you verify that VLAN port conf is not an issue, by testing connectivity on eth1 in a no-vlan envoronment ? or on untagged ports ?

  4. #4
    Join Date
    Apr 2006
    Beans
    41

    Re: Weird multi-home issue....

    Quote Originally Posted by koenn View Post
    what elica says.

    or

    Are you sure ? Did you verify that VLAN port conf is not an issue, by testing connectivity on eth1 in a no-vlan envoronment ? or on untagged ports ?
    VLAN config is verified, also moved the nic to a port that is currently used by one of my esxi servers... The server is operational in the port previously used by the NAS box and the NAS box is inoperable in the ports previously used by the esxi server.

    Watched the activity lights and it appears that traffic is still routed through eth0 when it should be going through eth1 for 192.168.100.xxx

    Very odd

  5. #5
    Join Date
    Apr 2006
    Beans
    41

    Re: Weird multi-home issue....

    Quote Originally Posted by tonyric View Post
    VLAN config is verified, also moved the nic to a port that is currently used by one of my esxi servers... The server is operational in the port previously used by the NAS box and the NAS box is inoperable in the ports previously used by the esxi server.

    Watched the activity lights and it appears that traffic is still routed through eth0 when it should be going through eth1 for 192.168.100.xxx

    Very odd
    Also, no firewall on the NAS box.

  6. #6
    Join Date
    Dec 2010
    Beans
    570
    Distro
    Ubuntu 12.04 Precise Pangolin

    Re: Weird multi-home issue....

    As long as you are not needing/wanting to access any other subnets connected to the 192.168.100.* through eth1 then not having a gateway is no issue.

    How are you doing the VLANs? Are you setting the default VLAN on the switch for the port or are you using VLANs on the Linux interface?

    What happens if you use the "-I' option to make ping use the eth1 interface? This will determine if it is an issue with routing determining what interface to use or there is an issue with communicating on the VLAN.

  7. #7
    Join Date
    Nov 2006
    Location
    Belgium
    Beans
    3,023
    Distro
    Ubuntu 10.04 Lucid Lynx

    Re: Weird multi-home issue....

    moved the nic to a port that is currently used by one of my esxi servers... The server is operational in the port previously used by the NAS box and the NAS box is inoperable in the ports previously used by the esxi server.
    That's not conclusive.
    If your esx nic and both switch ports are VLAN-"tagged" ( http://en.wikipedia.org/wiki/IEEE_802.1Q ) but your NAS nic isn't, that's exactly the expected result, at least on the HP Procurve switches a use.

    That's the reason hawkmage's question is relevant.

    What if you connect an other machine (a laptop, ...) -- with appropriate network config i.e. in the same subnet -- directly to the NAS, no switches involved. Can they ping each other ?

  8. #8
    Join Date
    Apr 2006
    Beans
    41

    Re: Weird multi-home issue....

    I re-verified the port settings, ports 1-8 are tagged vlan id 1 which is used for 192.168.0.xxx and 9-16 for vlan 100 used by 192.168.100.xxx. All ports are tagged correctly, also all ports on both vlans are allowed to accept tagged and untagged frames. There is no tagging happening on the nas box, the ESXi boxes are tagged as that's the default config for vmware. I will enable tagging on the NAS box and see if that fixes the issue, will note this thread for others if it works.

    NOTE: My vCenter server works as multi-homed (virtual machine), but this works of course because it is using the same physical nics as the ESXi servers on vlan 100 and both ESXi servers talk find on vlan 100 regardless of the port used that are configured for vlan 100. So, it looks like tagging is the issue.
    Last edited by tonyric; May 15th, 2012 at 05:38 AM.

  9. #9
    Join Date
    Dec 2010
    Beans
    570
    Distro
    Ubuntu 12.04 Precise Pangolin

    Re: Weird multi-home issue....

    I also run ESXi but I have 3 VLANs defined and trunked across 3 switches.

    Did you try the ping with the "-I" option to specify the eth1 interface to force it to use the proper interface?

    I assume that you are not using VLAN tagging on the Linux interface eth1 since normally a Debian VLAN interface is a virtual interface on top of a physical one and you would likely not be using the name eth1.

    Besides being a member of VLAN 100 are you having the switch tag untagged packets as VLAN 100 on ports 9-16? If not the switch will treat untagged packets as VLAN 1.

  10. #10
    Join Date
    Apr 2006
    Beans
    41

    Re: Weird multi-home issue....

    It was vlan tagging..... Simple fix:

    apt-get install vlan
    vi /etc/network/interfaces
    change eth1 to eth1.100
    Save file

    modprobe vlan
    ifup eth1.100

    all works

    add 8021q to /etc/modules

    Thanks everyone for pointing me in the right direction.

Bookmarks

Posting Permissions

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