[nycbug-talk] arp questions

[Pete Wright]

Okan Demirmen wrote:

>On Tue 2004.06.29 at 11:22 -0400, Pete Wright wrote:
>  
>
>>hey all,
>>   i've got a question regarding ARP on FreeBSD 4.10.  I have two 
>>machines (archiver and homer).  each machine has two interfaces, one on 
>>our 100Base-T LAN and another on our seperate Gig-E LAN.  Each interface 
>>is given a different name, so the gig-e interface for homer is 
>>gate-homer (don't ask why we use gate-hostname wierd SGI thing...).  
>>Anyway when I perform a large NFS copy via the "gate" interfaces is get 
>>this error in my syslog:
>>
>>arp: 192.169.3.10 is on em0 but got reply from xx:xx:xx:xx:xx:ee on fxp0
>>(let's pretend that's a valid MAC address, the last two bits are the 
>>only interesting part anyway).
>>    
>>
>
>uhm, are you positive that the two networks are not touching each
>other?  it looks like the two networks are bridged without
>spanning-tree. unless of course, you have say a sun box with multiple
>nics on the two netowrks, all with the same mac address; that works
>fine, but if other boxes are on the same networks in the same fashion, 
>you can run into arp issues.
>
>  
>

nope they are not touching.  the gig-e network does not even route to 
the internet, only the 100base-T one does.  each nic does have a 
different MAC address as well.  I forgot to mention that the NFS server 
in this case is a Debian gnu/linux box(2.4.26+XFS), the client is 
FreeBSD 4.10.  I am starting to suspect that there may be an issue with 
the Debian box, I do not run into this issue with any of my SGI or other 
BSD boxen on the gig-e network.

-p


-- 
~~~oO00Oo~~~
Pete Wright
email:  pete at nomadlogic.org
mobile: 917.415.9866
web:    www.nomadlogic.org/~pete