IPv6 Issues - Lost IPv6 routing (IPv4 NOT affected)

[jpatriarca]

Hi team,

I have a two node setup that, regularly, stop having Ipv6 routing configurations, but Ipv4 connectivity works as expected. Basically, anytime I detect that the issue are present, I go to the admin web and renew interfaces configuration or restart Networking service for each host. When this happens, IPv6 network is restored.

What I can see is that, using ip neighbour show command, each IPv6 router IP have STALE status or simply disappears, which don't happen in IPv4. After netowkring service restart, the status for this command seems the same but connectivity is restored.

All hosts and CT/VM have fixed IPv4 and IPv6 and the virtual machines don't lost connectivity at all. Sharing my configuration, and please ask if you need further information:

Interface config - Host #1

auto lo
iface lo inet loopback

iface enp86s0 inet manual
        mtu 9000

auto vmbr0
iface vmbr0 inet static
        address 192.168.1.21/24
        gateway 192.168.1.1
        bridge-ports enp86s0
        bridge-stp off
        bridge-fd 0
        mtu 9000

iface vmbr0 inet6 static
        address 2001:xxxx:xxxx:xxxx::21/64
        gateway 2001:xxxx:xxxx:xxxx::1

ip neighbour show command - Host #1

192.168.1.12 dev vmbr0 lladdr ea:25:4e:d0:c7:d3 STALE
192.168.1.1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 REACHABLE
192.168.1.22 dev vmbr0 lladdr 1c:69:7a:68:8d:93 REACHABLE
192.168.1.10 dev vmbr0 lladdr 00:11:32:6e:21:9d REACHABLE
192.168.1.53 dev vmbr0 lladdr ea:25:4e:d0:c7:d3 STALE
192.168.1.13 dev vmbr0 lladdr dc:a6:32:45:2f:9b REACHABLE
fe80::74ac:b9ff:fe35:16b2 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE
2001:xxxx:xxxx:xxxx::1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE
2001:xxxx:xxxx:xxxx::22 dev vmbr0 lladdr 1c:69:7a:68:8d:93 router REACHABLE
fe80::1e69:7aff:fe68:8d93 dev vmbr0 lladdr 1c:69:7a:68:8d:93 router STALE

ip neighbour show command (after networking service restart and IPv6 connectivity restored)- Host #1

192.168.1.12 dev vmbr0 lladdr ea:25:4e:d0:c7:d3 STALE
192.168.1.1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 REACHABLE
192.168.1.22 dev vmbr0 lladdr 1c:69:7a:68:8d:93 REACHABLE
192.168.1.10 dev vmbr0 lladdr 00:11:32:6e:21:9d REACHABLE
192.168.1.53 dev vmbr0 lladdr ea:25:4e:d0:c7:d3 STALE
192.168.1.13 dev vmbr0 lladdr dc:a6:32:45:2f:9b REACHABLE
fe80::74ac:b9ff:fe35:16b2 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE
2001:xxxx:xxxx:xxxx::1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE
2001:xxxx:xxxx:xxxx::22 dev vmbr0 lladdr 1c:69:7a:68:8d:93 router REACHABLE
fe80::1e69:7aff:fe68:8d93 dev vmbr0 lladdr 1c:69:7a:68:8d:93 router STALE

Interface config - Host #2

[CODE]auto lo
iface lo inet loopback

iface eno1 inet manual
        mtu 9000

auto vmbr0
iface vmbr0 inet static
        address 192.168.1.22/24
        gateway 192.168.1.1
        bridge-ports eno1
        bridge-stp off
        bridge-fd 0
        mtu 9000

iface vmbr0 inet6 static
        address 2001:xxxx:xxxx:xxxx::22/64
        gateway 2001:xxxx:xxxx:xxxx::1

ip neighbour show command - Host #2

192.168.1.13 dev vmbr0 lladdr dc:a6:32:45:2f:9b REACHABLE
192.168.1.1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 REACHABLE
192.168.1.21 dev vmbr0 lladdr 48:21:0b:34:d2:de REACHABLE
192.168.1.10 dev vmbr0 lladdr 00:11:32:6e:21:9d REACHABLE
2001:xxxx:xxxx:xxxx::21 dev vmbr0 lladdr 48:21:0b:34:d2:de REACHABLE
fe80::74ac:b9ff:fe35:16b2 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE

ip neighbour show command (after networking service restart and IPv6 connectivity restored)- Host #2

192.168.1.13 dev vmbr0 lladdr dc:a6:32:45:2f:9b REACHABLE
192.168.1.1 dev vmbr0 lladdr 76:ac:b9:35:16:b2 REACHABLE
192.168.1.21 dev vmbr0 lladdr 48:21:0b:34:d2:de REACHABLE
192.168.1.10 dev vmbr0 lladdr 00:11:32:6e:21:9d REACHABLE
fe80::4a21:bff:fe34:d2de dev vmbr0 lladdr 48:21:0b:34:d2:de router STALE
2001:xxxx:xxxx:xxxx::21 dev vmbr0 lladdr 48:21:0b:34:d2:de router REACHABLE
fe80::74ac:b9ff:fe35:16b2 dev vmbr0 lladdr 76:ac:b9:35:16:b2 router STALE

Request your help to understand the root cause and what information might be missing? I don't see this behavior in any other device in my network, with DHCPv6 or fixed IP.

Also, VLAN aware settings are disabled since I've activated after the VM/CT creation and all CT/VM created after the change, lost networking connectivity (might a host restart be missing???).

Thank you

 

[Pedi]

Hi,
i have the same problem, after x-minutes my ipv6 connection dont work anymore but IPV4 work. I have to restart my Root Server 1-2 Times before it works again for x-minutes.

My config is from the base like yours, i also have:
post-up echo 1 > /proc/sys/net/ipv6/conf/all/forwarding
post-up echo 1 > /proc/sys/net/ipv6/conf/default/forwarding

Have you found any solution? (Please share, dont say only "Yes" )
Thanks

 

[jpatriarca]

Unfortunately not at all. From time to time it happens. I need to restart networking service in each host to restore ipv6 connectivity.

Also, during some weeks I’ve detected if I have more than one network interface with a different vlan the issue happens more often, like minutes, as you describe. But for the moment I’m not using it.

The big issue I have when this happens is that I cannot access to the host portal when Ipv6 is used for connectivity.

 

[encryptedserver]

Are you allowing DHCPv6 in your firewall, if not then IPv6 stops working after like 20 minutes.

Try adding net.ipv6.conf.vmbr0.accept_ra=2 in your /etc/sysctl.conf and reboot your proxmox and see if the problem persists.

 

[AndreasLink]

Did anyone successfully solve this problem? I am still (version PVE 8.2.7) in that situation and IPv6 stays happy for around 20 to 30 minutes and then just stops routing aka forwarding from the VMs. I can usually solve this by rebooting a client VM (Debian and Ubuntu tested).

So, in this case, I have a server in a hosting center in the internet running Proxmox. I have a static /64 and I have assigned to two of the VMs each a /80 (of that /64). /proc/sys/net/ipv6/conf/all/forwarding is activated as well as for vmbr0. It all looks correct, I boot and I have a ping6 from my local machine (here at home) to my server in the cloud as well as into the VM. I can also use IPv6 from the VM and all is as it should.

An even when I let the ping just go, it suddenly stops after the ~20 - 30 mins and I have no clue what kicks in here. But I can still ping Proxmox, but none of the VMs anymore.

So, this seems to be the same problem still, I have tried a lot already, but nothing helped. Looking for ideas.

 

[jpatriarca]

Did anyone successfully solve this problem? I am still (version PVE 8.2.7) in that situation and IPv6 stays happy for around 20 to 30 minutes and then just stops routing aka forwarding from the VMs. I can usually solve this by rebooting a client VM (Debian and Ubuntu tested).

So, in this case, I have a server in a hosting center in the internet running Proxmox. I have a static /64 and I have assigned to two of the VMs each a /80 (of that /64). /proc/sys/net/ipv6/conf/all/forwarding is activated as well as for vmbr0. It all looks correct, I boot and I have a ping6 from my local machine (here at home) to my server in the cloud as well as into the VM. I can also use IPv6 from the VM and all is as it should.

An even when I let the ping just go, it suddenly stops after the ~20 - 30 mins and I have no clue what kicks in here. But I can still ping Proxmox, but none of the VMs anymore.

So, this seems to be the same problem still, I have tried a lot already, but nothing helped. Looking for ideas.

I can tell you that your report is not the same as I have.

My issue is related with IPv6 traffic between hosts. Any container or VM that I have in my hosts, even with IPv6 configured works perfectly.

At this moment, both hosts have only IPv4 on vmbr0 interface and each ct/vm have their own IPv6, fixed or via DHCP6/RA and works perfectly.

 

[AndreasLink]

I can tell you that your report is not the same as I have.

My issue is related with IPv6 traffic between hosts. Any container or VM that I have in my hosts, even with IPv6 configured works perfectly.

At this moment, both hosts have only IPv4 on vmbr0 interface and each ct/vm have their own IPv6, fixed or via DHCP6/RA and works perfectly.

OK, so you had no problems talking to the web via IPv6 from the guests?

I just found another discussion, which is exactly adressing and explaing the problem, which also seems to be a bit hoster related, when I got that right. There the MAC address used triggers the issue, maybe, this is also relevant for you?

https://serverfault.com/questions/916309/ipv6-connectivity-lost-on-kvm-guest-after-20-minutes

But I measured it now and it hits exactly the 20 minutes, so then I guess, I will do some more research on the Hetzner related issue.