Comware

 View Only
Expand all | Collapse all

Server 2012 Nic Team HP 2920 problem

This thread has been viewed 0 times
  • 1.  Server 2012 Nic Team HP 2920 problem

    Posted Oct 08, 2013 10:32 AM

    Hi!

     

    I have a hyper-v 2012 cluster with two hosts. Every host has 4 nic ports. On each host the 4 nics are in a switch independent network team (Team Server A and Team Server B). On top of each team is a vSwitch with the cluster, management and livemigration network. 

     

    The physical network configuration is as follows. There are 3 switches (A, B and C).

     

    Switch A = HP 2920-24G

    Switch B = HP 2920-24G

    Switch C = Netgear

     

    Ports 1 and 2 of Team Server A are connected to ports 1 and 2 of Switch A

    Ports 3 and 4 of Team Server A are connected to ports 1 and 2 of Switch B

    Ports 1 and 2 of Team Server B are connected to ports 3 and 4 of Switch A

    Ports 3 and 4 of Team Server A are connected to ports 3 and 4 of Switch B

     

    The crossover cabling is for fault tolerance if Switch A or Switch B dies.

     

    Port 24 and 23  on Switch A and Switch B are trunked uplinks to Switch C which is the production network.

    Spanning Tree is active on all switches

     

    If I connect the network teams in the crossover configuration the whole network goes down. Not even a ping to any PC or Server is possible.  If all ports of the network teams are attached to one switch (A or B) everything is ok.

     

    A guy on Microsoft Tech Net verified my Hyper-V configuration and said, that it must have something to do with the switches. He mentioned "switchport nonegotiate".

     

    Is there anything wrong with my cabling or do I have to configure something on the switches to get the fault tolerant network team working?

     

    Hope anybody can help!

     

    Thank's in advance

     

    Madmax


    #NIC


  • 2.  RE: Server 2012 Nic Team HP 2920 problem

    Posted Oct 08, 2013 07:21 PM

    I don't see any obvious reason why connecting it all up the way you plan should bring the whole network down.

     

    A couple of things:

     

    1/ Have you reduced the spanning tree priority of Switch C to 0 (or something lower than the default 32k)?

     

    2/ Have you got the trunks from Switch C up to A & B each configured as a working LACP trunk?

     

    3/ Have you got the trunks from each switch (A & B) to each Host configured as working LACP trunks?