AIX

AIX

Connect with fellow AIX users and experts to gain knowledge, share insights, and solve problems.


#Power
 View Only
  • 1.  AIXIF_ARP_DUP_ADDR on bladecenter woes

    Posted Wed March 12, 2008 12:51 PM

    Originally posted by: Yanuly


    Hi, I have a J21 VIO Server which is complaining of AIXIF_ARP_DUP_ADDR (duplicate address on the net), the supposed culprit is 0014 5E3C CC30 which is the MAC of the selfsame BladeCenter. I'm guessing this has to do with the Etherchannel the VIO configures... Any hints? I couldn't find anything on the IBM site-manuals-papers...
    EtherChannel / Link Aggregation ent2
    Parent Adapter ent7
    EtherChannel / Link Aggregation Adapters ent0 +
    Backup Adapter ent1 +
    Automatically Recover to Main Channel yes +
    Perform Lossless Failover After Ping Failure yes +
    Enable Alternate Address no +
    Alternate Address 0x000000000000 +
    Enable Gigabit Ethernet Jumbo Frames no +
    Mode standard +
    Hash Mode default +
    Internet Address to Ping [0]
    Number of Retries [3] +#
    Retry Timeout (sec) [1] +#
    #AIX-Forum


  • 2.  Re: AIXIF_ARP_DUP_ADDR on bladecenter woes

    Posted Thu March 13, 2008 01:02 PM

    Originally posted by: eichher


    Do you have a Intel blade with Broadcom ethernet adapter running somewhere in your bladecenter, if yes points the MAC address given in your errpt entry to that blade??? If this is true you should destroy the adapter teaming on that machine and the DUP_ADDR message should be gone.
    #AIX-Forum


  • 3.  Re: AIXIF_ARP_DUP_ADDR on bladecenter woes

    Posted Thu March 13, 2008 06:35 PM

    Originally posted by: Yanuly


    Well, yes, I've just checked and the conflicting intel blade does Teaming. Problem is that we wouldn't want to destroy that as the volume it handles uses that extra capacity teaming offers.
    While reading some docs I wonder if I can replace the pseries Etherchannel with Link Aggregation, would the blade switch support that? The Help for etherchannel/LA says:

    The "8023ad" mode enables the Link Aggregation Control Protocol
    (LACP) to negotiate the adapters in the link aggregation with a
    LACP-enabled switch.

    Why haven't I tried this? The server is already in use and since etherchannel seemed harmless enough I hope we can solve this in a more definitive and knowledgeable way, sooner and final that is.
    Thanks!
    #AIX-Forum


  • 4.  Re: AIXIF_ARP_DUP_ADDR on bladecenter woes

    Posted Fri March 14, 2008 04:54 AM

    Originally posted by: eichher


    I don't know the behavior if you use link aggregation instead of failover mode. We opened an IBM AIX call for the AIXIF_ARP_DUP_ADDR message , here is the answer:
    the testing from IBM Japan had obvious identified the source of strange packet from the IP trace.
    In that case, the strange packet was being sent out from the window box. AIX sent gratuitous arp, the window box replied with the sender/target ip were both same IP as AIX and sender mac from window and target mac from AIX. The answer IBM got from Broadcom Support: This is a known issue with BASP, we are planning to qualify the fix by mid Descember.
    So check if there is a new broadcom ethernet driver for your INTEL blade, perhaps this will solve the problem.
    #AIX-Forum


  • 5.  Re: AIXIF_ARP_DUP_ADDR on bladecenter woes

    Posted Wed March 19, 2008 12:32 PM

    Originally posted by: Yanuly


    Thanks! What solved the issue was just to switch to Link Aggregation on the Etherchannel/Link Agg config menu and reboot:
    Mode 8023ad +
    #AIX-Forum