docs/159374: Difficulty following guide on configuring link aggregation

Benjamin Kaduk kaduk at MIT.EDU
Thu Aug 4 20:28:53 UTC 2011

Dear Karli,

On Tue, 2 Aug 2011, Karli SjXberg wrote:

> last week, it was the first time for me setting up a lagg interface with 
> a cisco switch and had this page from the handbook as a guide:
> Link Aggregation and Failover
> I had troubles getting everything to work because the lagg interface 
> always showed "no carrier" as status and couldn?t get an address. After 
> much googling, I found the reason for this is because you have make the 
> laggports "up" before you create the lagg interface.
> The rest went off without a hitch and it was so cool to see everything 
> "just work". You really got double the perfomance and perfect failover!
> If this had been explained in the handbook, it would have saved me half a day?s worth of troubleshooting.
> Correct approach:
> # ifconfig em0 up
> # ifconfig em1 up
> # ifconfig lagg0 create
> # ifconfig lagg0 up laggproto lacp laggport em0 laggport em1

This is standard behavior for manually configuring network interfaces -- 
the explicit setting of the interface as 'up' is needed when no other 
information about the interface is given (e.g. in rc.conf or 
wpa_supplicant.conf).  So I do not think it is really appropriate to give 
particular emphasis to the need to "up" the interface, in this particular 
location in the text.
It should be fine to include an explicit example of doing so for the LACP 
case, though.  (Note that a later, non-LACP, example on that page does 
include the 'up' stage.)
It may also be needed to mention the general need for setting interfaces 
as "up" elsewhere in the handbook, but I do not have time to check that 
right now.

> /etc/rc.conf:
> ifconfig_em0="up"
> ifconfig_em1="up"
> cloned_interfaces="lagg0"
> ifconfig_lagg0="up laggproto lacp laggport em0 laggport em1 DHCP"

Is the "up" really necessary for ifconfig_lagg0, here?  I could mostly 
believe that it is needed for em[01], but my understanding is that it 
would not be needed for lagg0; the very last example on the page you link 
does not have it for the failover case, as well.  If you could check and 
report back, that would be helpful.


Ben Kaduk

> I think it would make things much easier for other people in that 
> situation, if this was explained on that page. How does that sound?
> /Sebulon
>> How-To-Repeat:
>> Fix:
>> Release-Note:
>> Audit-Trail:
>> Unformatted:
> _______________________________________________
> freebsd-doc at mailing list
> To unsubscribe, send any mail to "freebsd-doc-unsubscribe at"

More information about the freebsd-doc mailing list