Container loosing network connection daily

[BoxSon]

Hi,

using ProxMox now for one year without any issues. From one day to another one of my containers is loosing network connectivity one a day. It's not always loosing the connectivity at the same time.
The container runs Debian GNU/Linux 11 (bullseye) (Openhabian distro). The container ran without any issues for around one year.
I can not find any log entries within the container logs, only that there was a change in network connection.

The container itself seems to be fine, i can open the console in PVE for the container but the only way to have the network connection working again is to restart the container.

There is nothing special about the container, worked for a long time with this configuration.
There are two other containers also runnig debian bullseye at the same PVE node. Both run flawless and have no networking issues.

As i'm more or less new to the PVE internals, maybe someone can give me a hint where to start analyzing the root caus?

 

[nunner]

Hi,

what does the network configuration look like for this container? My first guess would be that you have duplicate IPs somewhere on the network, have you checked for that?

 

[BoxSon]

Hi Leo,

The container is in a VLAN. This is how it has worked since the beginning.
The IP address for the container is set statically in the DHCP server, and I have also set the IP address statically within the container.

There are only a few devices in the network segment, and all of them have statically set IP addresses in the DHCP server.

If this would be a dhcp problem or a problem with duplicate IP-Adresses, the problem should show up more or less right 24 hours after the last reboot of the container, correct?

Regards

 

[nunner]

The IP address for the container is set statically in the DHCP server, and I have also set the IP address statically within the container.

I'm not sure if I follow here. You have the IPs set statically inside the container, and the addresses are then excluded from the pool on the server?

I can not find any log entries within the container logs, only that there was a change in network connection.

Could you still post these logs (with regards to the change in network connection)? Do you remember if you did any major upgrades/changes in hosts/[…] before this started happening?

 

[BoxSon]

Just had the same situation. The Container dropped network connection today at 01:38pm

Syslog:

Dec  7 13:08:36 openhabian avahi-daemon[106]: Withdrawing address record for 192.168.3.241 on eth0.
Dec  7 13:08:36 openhabian avahi-daemon[106]: Leaving mDNS multicast group on interface eth0.IPv4 with address 192.168
Dec  7 13:08:36 openhabian avahi-daemon[106]: Interface eth0.IPv4 no longer relevant for mDNS.

Thats the only information i get within the syslog.

I didn't had any major upgrades nor changes to the container system so far. Also the pve hasn't been updated around the moment that happened first.

I'm not sure if I follow here. You have the IPs set statically inside the container, and the addresses are then excluded from the pool on the server?

My network is managed by a ubiquiti dream machine. This device offers DHCP for all networks. At the dream machine you can define a static ip address for a known device (by mac address i guess).

The network configuration of the container linux system in /etc/network/interfaces is as follows:

auto eth0
iface eth0 inet static
address 192.168.3.241
netmask 255.255.255.0
gateway 192.168.3.1
dns-nameservers 192.168.3.1

 

[nunner]

Are your static IPs defined inside the DHCP ranges? I don't have any experience with ubiquiti devices and how they would handle that situation, but maybe that's something that could cause issues? Are there maybe any logs on your router from around the time that this happens?

 

[Graefe]

Same here. All containers loose network after 24h, if they are connected with DHCP. Workaround for me: make IPv6 static an leave IPv6 address empty.

 

[psychozoic]

and here also same.

containers are based on debian12, in dhcp reservation addresses provided, running from os dhclient helps to get ip.

 

[a575606]

I'm going to throw my hat in the ring too.. I have 4 lxc containers, all based on the same image, Debian 12, all dhcp with static leases set on the DHCP server. I haven't tried using a different distro, so I am not sure if it's a Debian or an LXC issue, or perhaps it's related to ifupdown. However it definitely seems an issue... All 4 containers lose network connectivity daily, i'm assuming when the DHCP lease expires. I'm trying to figure out how to switch over to systemd networkd and disable ifupdown, but if that doesn't solve the problem I may just axe the containers and move to a vm. One of the containers is running dns, so it's awfully inconvenient when it goes offline.

If it makes any difference, my DHCP is running on my main router/firewall, a Mikrotik device.