[pve-devel] LXC IPv6 not set after reboot

Wolfgang Bumiller w.bumiller at proxmox.com
Thu Nov 16 14:24:06 CET 2017


On Wed, Nov 15, 2017 at 01:03:30AM +0100, Bastian Sebode wrote:
> Hello,
> 
> I expierience a bug with IPv6 in Containers on PVE 5.1-36. When I set
> IPv6 address and IPv6 gateway for the container through the Webinterface
> it gets set correctly, but when i reboot the container it doesn't get
> set again, although it remains in the container configuration.
> 
> When i change the IPv6 it gets set in the container again, also for gateway.
> 
> So after a container reboot i change IPv6 address and IPv6 gateway twice
> to have the correct configuration being present again.

Rebooting a container does the same as stop+start, and the system in
the container itself is responsible for setting up the network, all we
do is write the /etc/network/interfaces file in between.
And I can't reproduce this. The only thing I could imagine is that DAD
(duplicate address detection) triggers - either because the address is
used in multiple places, or something causes the network namespace to
linger (but this should cause naming conflicts with the veth devices and
the container should refuse to start at all...).




More information about the pve-devel mailing list