Carp: checksum failed on em0

Ruben van der Zwan freebsd-net at m0z.net
Thu Mar 9 14:51:08 UTC 2006


Your remark about the switch being the problem made me think :)
I captured the packets with tcpdump which resulted in some announces 
from an IP transit provider. The len of 20 was something I found also in 
that packet, so I'll send the logs to the provider, asking them to look 
into it. I don't believe it's something carp is causing, it's just 
reporting about the packets...

Anyhow, thanks for your time!

Gleb Smirnoff wrote:
> On Wed, Mar 08, 2006 at 11:13:08PM +0100, Ruben van der Zwan wrote:
> R> I have 2 routers running carp:
> R> 
> R> router1:
> R>    ifconfig_carp0="vhid 1 pass hYYgtsRE 192.168.1.1/24"
> R>    ifconfig_carp0_alias0="vhid 1 pass hYYgtsRE 192.168.2.1/24"
> R> router2:
> R>    ifconfig_carp0="vhid 1 advskew 100 pass hYYgtsRE 192.168.1.1/24"
> R>    ifconfig_carp0_alias0="vhid 1 advskew 100 pass hYYgtsRE 192.168.2.1/24"
> R> 
> R> There are vlan's present on the routers, but the 192.168.1.2/2.2 for 
> R> router1 and 192.168.1.3/2.3 for router2 IP addresses are present on an 
> R> em0 interface.
> R> 
> R> /var/log/messages (on both router1 and router2) is flooded with these 
> R> messages:
> R>    Mar  8 21:53:58 router1 kernel: carp_input: checksum failed on em0
> R>    Mar  8 21:53:58 router1 kernel: carp_input: received len 20 < 
> R> sizeof(struct carp_header)
> R> once every second...
>
> It looks like you are receiving bad packets on wire. Can you please run
> tcpdump and capture the CARP announces. How do they look like?
>
> You can also try to change the switch, the NICs and see whether it helps.
>
>
>   



More information about the freebsd-net mailing list