[SOLVED] Static IPv6 - Lost route after a while

[BJ78945]

Hi,

I have a strange problem. I use a server at ovh with ovh proxmox 4 template. IPv6 is working. I defined static entries as described in the wiki. But after a while i cannot access the vm via ipv6. When i make a ping from the vm to any ipv6 address, i get the message network is unreachable. So i try to debug my problem. I see after a rebbot i get default routes via ra and with this it works. When this routes expire, ipv6 didn't work anymore. I cannot see the static default route but if I manual add this, I get the error that this entry allready exists.

VM:
(Proxmox GUI Network settings see attachment)

iface eth0 inet6 static
        address XXX
        netmask 64
        post-up ip route add XXXff:ff:ff:ff:ff dev eth0
        post-up ip route add default via XXXff:ff:ff:ff:ff dev eth0
        pre-down ip route del default via XXXff:ff:ff:ff:ff dev eth0
        pre-down ip route del XXXff:ff:ff:ff:ff dev eth0

ip -6 route
XXX::/64 dev eth0  proto kernel  metric 256
XXXff:ff:ff:ff:ff dev eth0  metric 1024
XXX::/56 dev eth0  proto kernel  metric 256  expires 2591042sec
fe80::/64 dev eth0  proto kernel  metric 256
fe80::/64 dev eth1  proto kernel  metric 256
default via fe80::XX:af00 dev eth0  proto ra  metric 1024  expires 842sec hoplimit 64
default via fe80::XX:df80 dev eth0  proto ra  metric 1024  expires 842sec hoplimit 64

Host:

iface eth0 inet6 static
        address XXX
        netmask 64
        post-up ip route add XXXff:ff:ff:ff:ff dev eth0
        post-up ip route add default via XXXff:ff:ff:ff:ff dev eth0
        pre-down ip route del default via XXXff:ff:ff:ff:ff dev eth0
        pre-down ip route del XXXff:ff:ff:ff:ff dev eth0

ip -6 route
XXX::/64 dev vmbr0  proto kernel  metric 256  pref medium
XXXff:ff:ff:ff:ff dev vmbr0  metric 1024  pref medium
fe80::/64 dev vmbr0  proto kernel  metric 256  pref medium
fe80::/64 dev tap151020097i0  proto kernel  metric 256  pref medium
fe80::/64 dev fwbr151020105i0  proto kernel  metric 256  pref medium
...
default via XXXff:ff:ff:ff:ff dev vmbr0  metric 1024  pref medium

EDIT:
If the ra default route expires, i can add the regulare route with "ip route add default via XXXff:ff:ff:ff:ff dev eth0" and then ipv6 works. I think the problem is the following: If a vm starts, everything get configred as described in interfaces config but the vm get also ra routes (perhaps from the host??). When the routes expires, alls default routes get deleted, so i can add the real default route again. But this is no solution beacause it dosn't survice a reboot. My vm uses debian from the proxmox images.

 

[wbumiller]

So there's something in the network sending router advertisement packets without renewing the routes after they expire? You'll want to fix your router and/or ignore router-advertisements on the container's interface.
(net.ipv6.conf.eth0.accept_ra=0)

 

[BJ78945]

Hi,

I tried net.ipv6.conf.eth0.accept_ra=0 but this didn't help. But the problem is gone. Perhaps there was a mistake in the network.

Thank you.

 

[voja]

I have a Soyoustart (OVH brand) server with Proxmox 4 installed. I had the same problem, default route got lost short after boot. I saw link-local gateways in the route command.

Setting net.ipv6.conf.eth0.accept_ra=0 inside the container seems to have fixed the issue with the OVH ra message. I really hope the error is gone now, it drove me crazy.

Is it possible to set these values for new containers per default?

Regards
voja

 

[BJ78945]

Hi voja,

my problem appears again. How about you? Is your fix still working?

 

[voja]

Hi,

my containers still all have IPv6 connectivity.
I added the file /etc/sysctl.d/ipv6.conf to all my Debian/Ubuntu containers with the content:

net.ipv6.conf.eth0.accept_ra=0

Then I rebooted all containers and checked the value to be set correct.

This solved my problem since ~23 hours. Without the entry the connectivity was lost within 15-30 minutes or did not work at all after rebooting the container. No problems since.

Regards
voja