On Fri, May 25, 2007 at 07:00:14PM +0100, Stuart Henderson announced:
quoted text
> If they don't all change state together, the most likely reason is a
> mismatching set of addresses on the carp interfaces. Note there is a
> bug fixed in sys/netinet/ip_carp.c 1.135 re: adjusting addresses on
> a running interface. If you ran into it, you'll see ipv6 proto 112
> packets on tcpdump on the parent interface.
> 
> -stable candidate, anyone?
 
The change in 1.135 was put in -stable a few weeks ago looks like.

-- 
am9obkBzY2FyZm9uZS5uZXR8fGowQGNveC5uZXQ=