bond + bridge not routing ipv6 correctly

[yougotborked]

Yesterday I migrated my home lab (6 vms, 2 esx servers), to proxmox using the clonezilla method. It all went well once I realized that I kept accidentally selecting the 16gb flashdrive proxmox is installed on as my vm's hard disk's datastore. Everything was working fine before, but my license key had expired a long time ago and I wanted live migration. my configuration is as follows:

eth0 wan interface -> vmbr0 -> pfsense wan
pfsense lan -> vmbr1 -> other vms and bond0 -> eth1,2,3 to switch

I started realizing after setting the network using the gui, that my ipv4 AND ipv6 traffic was really flaky, packets would drop, and in general things were really slow. I also noticed that in my syslog was the following, repeating itself thousands of times per minute

Nov  3 18:11:03 kamino kernel: bond0: received packet with own address as source address

I started searching around the forums here for people experiencing the same problem, and I found a couple threads which provided a solution to add a couple lines to your bridge config (indicated with ## below). I did that and it fixed my ipv4 problems immediately. All that traffic works great.

I then noticed that I was not getting dhcp6 leases, so I tried various ping scenarios, and tentatively determined that I cannot get ipv6 traffic through my bridge. This is my current config, does anyone have suggestions to get this to work?

# network interface settings
auto lo
iface lo inet loopback

iface eth0 inet manual

iface eth1 inet manual

iface eth2 inet manual

iface eth3 inet manual

auto bond0
iface bond0 inet manual
        slaves eth1 eth2 eth3
        bond_miimon 100
        bond_mode balance-xor

auto vmbr1
iface vmbr1 inet static
        address  10.1.1.8
        netmask  255.255.255.0
        gateway  10.1.1.1
        bridge_ports bond0
        bridge_stp on
        bridge_maxwait 0 ##
        bridge_maxage 0 ##
        bridge_fd 0
        bridge_ageing 0 ##

auto vmbr0
iface vmbr0 inet manual
        bridge_ports eth0
        bridge_stp off
        bridge_fd 0

 

[fireon]

I work on the same problem with Ubuntu. I don't know, but i think this feature use nobody. Every little network has an layer2 switch and with this you can use 802.3ad. With this i think it is not a problem... but... i've never tested this with IPV6/4 before. With IPV4 it works fine. Have a hand full servers running with 4nics+

 

[athompso]

Short answer: don't use vmbr with balance-xor. Most of the other modes will work fine. LACP (aka 802.3ad) definitely does work correctly, but that requires configuration in your next-hop (hardware) switch. Active-passive will definitely work correctly. Balance-arp *should* but I've never tested it.
-Adam