[jrigler]

In the past year, I have seen this error suddenly crop up in my error report:

AIXIF_ARP_DUP_ADDR

It inevitably points to a MAC address of some other machine on our network
that is not trying to use my ip address. I seem to only see this when:

1. Using an etherchannel with one primary and one backup adapter going to
two switches.

2. Using the arp=yes option

I believe that since etherchannel seems to broadcast one MAC address to two different ISLd switches, that this is where the message comes from, without the backup adapter, I don't see this.

Has anyone else seen this? Any ideas?

[Frans]

This error is because a DUPLICATE IP ADDRESS DETECTED IN THE NET


Now depending on your setup there are a few reasons this can come up
· Duplicat ip address on the network check the address
· You have problem on your Oracle Rac Check the logs
· If you have HACMP there is a problem with the Cluster Check the logs

[jrigler]

Its a fresh install in this case, no HA or Oracle yet. I only see it when I have an etherchannel and specify the backup option. I think that something has changed in the network or vlan configuration that makes the two halves of the etherchannel suddenly see each other and get upset. I am going to try a true etherchannel again rather than a primary/backup.

[Administrator]

You can't configure an etherchannel across multiple switches.

[duke900ssd]

You can use different switches if you are using Network Interface Backup.
If you are using link aggregation then it gets more complicated...

NIB Link:
IBM Systems Information Center

[jrigler]

This is what we do. I am now wondering if this is message is fixed with more recent device drivers or microcode.

[camutia]

Hi, have you been able to resolve this one? I am facing the same situation with one of my implementations and I was wondering how to correct this...

[dthacker]

I've got this problem too. The VAR blames the network guys and the network guys blame the var. Any good way to diagnose and resolve?

TIA Dave

[stevosoft]

We have had this for a while so we placed a call with IBM support who advised us it was a known problem in networks containing MS windows boxes of a certain version. I can't remember the full details but basically it came down to not being a problem with the AIX boxes. If you give IBM a call they may be able to give you some extra info.