On 2012-04-25, Matt Hamilton <ma...@netsight.co.uk> wrote: > BARDOU Pierre <bardou.p <at> mipih.fr> writes: > >> >> Hello, >> >> I have dozens of CARP interfaces over VLAN interfaces over LACP trunk >> interfaces over physical EM/BGE/BNX. Carp is in multicast mode, multicast >> routing is disabled. Works like a charm with various OpenBSD versions since >> 4.4 to 5.0. > > OK, that is good to know. Are you using i386 or amd64? I'm wondering if > somehow that is a factor here? > >> I can give you my hostname.if if that helps... > > Yes, that would be very useful. Can you email to matth at netsight.co.uk. > > I'm about to start setting up ospfd on these hosts, which also uses > multicast so it will be interesting to see if that now also fails due to > multicast being filtered out somewhere.
You can try something simple to test multicast; ping6 -w ff02::1%vlan119 you should see name responses from the local machine and other boxes on the same vlan. I setup carp-on-vlan-on-trunk-on-bnx0/1 on an R210-II running 5.1 the other day, no trouble. In this case they're webservers so I didn't set net.inet.ip.forwarding in sysctl.conf and i'm using ip balancing rather than simple carp failover. hostname.carp82 carpdev vlan82 pass bleh advbase 0 carpnodes 74:32,75:128 balancing ip xxx.xxx.xxx.xxx/32 hostname.vlan82 vlandev trunk0 vlan 82 xxx.xxx.xxx.yyy/28 hostname.trunk0 trunkproto failover trunkport bnx0 trunkport bnx1 up hostname.bnx0, hostname.bnx1 up (using "advbase 0" allows for much faster failover, you really want 5.1 if you want to use that, it was added in 5.0 but the timer was a bit over-sensitive). erm, do you have a default route configured? I haven't tested, but it wouldn't be a big surprise if that resulted in behaviour like you're seeing.
