LXC secondary NIC

[SamTzu]

I have a weird problem with LXC containers. I can't seem to get up the secondary VM NIC IP's with out restarting networking inside the container. Any ideas? Everything seems to be ok but the secondary IP's never "take hold" without /etc/init.d/networking restart inside the VM.

 

[fireon]

NICs are not managed in the container, they are managed in the PVE Webinterface.

 

[SamTzu]

I know that. That's why this is weird.

 

[fireon]

Strange, what system? Have you tested DHCP and static?

 

[SamTzu]

I'm running Ubuntu 16 LXC container with 2 NIC's. One on public net and one on the private net with 192.168.14.z address. Both NIC's have static IP's but only the public one comes up after VM is started. I have to manually go in the VM and restart the networking before the second IP comes UP.

Before...

44: eth1@if45: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 8a:33:50:52:97:9c brd ff:ff:ff:ff:ff:ff link-netnsid 0

After...

44: eth1@if45: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether 8a:33:50:52:97:9c brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 192.168.14.14/24 brd 192.168.14.255 scope global eth1
       valid_lft forever preferred_lft forever
    inet6 fe80::8833:50ff:fe52:979c/64 scope link
       valid_lft forever preferred_lft forever

 

[SamTzu]

I solved this by adding this to cron : ) @reboot "/etc/init.d/networking restart"

 

[daemaz]

I don't have that issue on my containers that have two NICs, but all of my containers that have two NICs are mostly CentOS with some Debian (my Ubuntu ones only have one NIC). I have two tables in /etc/iproute2/rt_tables (one for each interface) with pre-up commands for routes and rules on each interface. Both interfaces come up but sometimes the service on that container fails. My CentOS container that hosts HAProxy will start successfully on boot about 50% of the time; my CentOS container for nginx will start successfully 0%. Once I go in and "systemctl restart <service>" then everything is good to go.

 

[SamTzu]

Seems like a systematic problem with the containers. Lets hope future upgrades will fix the problems.

 

[SamTzu]

Today we lost secondary interface all by it self while VM was running normally. /etc/networking/interfaces restart fixed the problem but this is getting kind of silly. Any ideas what could be causing these?

The primary public Interface always works.

 

[lordcheeto]

Issue in my case was the eth0 ipv6 being set to DHCP, which I don't have set up. eth0 ipv4 was starting, then it was timing out on ipv6, never moving on to bringing eth1 up. Disabled ipv6, et voila.

 

[SamTzu]

That is a good rule of thumb. Always blame IPv6.