Issue with VM network

This is continuation to the network manager issues (Wicked does not start automatically after update - #10 by paju-21) that I cannot figure out.

Before the change to network manager it was possible to ping the VM normally from other VLANs normally. Also, connect to email port (993) bur now this has stopped. I can connect from outside and also from VM host (in same network) but not from different VLAN. This used to work just fine.

Checked with another VM which has same NIC configuration for the VM and there’s no problems. I can ping this VM from different VLAN without problems. Added sshd and can connect to the VM from host and other VLAN just fine as well.

So, route to the VM seems to be fine but for some reason after the network manager change VM is blocking access. journalctl does not provide any logs. Firewall on VM is not running.

Any tips how to start solving this problem?

Try switching to systemd-network. I have no idea if using a VM makes any difference, but I have 0 installations using Wicked, 1 installation using NM (on laptop), and well over 100 openSUSE installations using systemd-network, which BTW is not supported by YaST, plus another hundred or so in distros other than openSUSE. /etc/resolv.conf I create manually. Like much about systemd, it’s simple, and reliable. NIC config goes in /etc/systemd/network/, where on Leap host from which I write I configure file eth0.network with:

[Match]
PermanentMACAddress=

[Network]
Address=
DNS=
Gateway=
IPv6AcceptRA=
LinkLocalAddressing=

Thanks for the 3rd alternative. Before jumping to this one, I’d really like to understand what is going on with the current approach using network manager. Wicked seems to be working but not network manager (others are using Wicked).

I guess the config name should be enp1s0.network instead etc0.network since the device name is enp1s0, right?

That’s my understanding. I like old-fashioned NIC naming, so my Grubs contain net.ifnames=0.

This tells us nothing. We have no idea how your network is organized, what are these VLANs (assuming they are VLANs and not something you decided to call so), how they are interconnected etc etc.

There is firewall on host to start with. At least, usually.

Start with sopping firewall everywhere.

They are VLANs (vlan101, vlan102, vlan103, vlan104, vlan105 - segmented network) so please stop this as this does not help at all - just annoys.

Firewall is not running on host. Only on the router and couple of Windows clients (just in case). VMs are not running firewall either.

Verified from the router that tcp communication (like port 993) tries to go to vlan101 from vlan102 but it does not progress. Communication from outside and directly from the same vlan works but not from other vlan. But as mentioned, this is not the case with other VM in same vlan101 with Wicked network management. Works just find; ping, ssh (no other service is currently there).

Changed to systemd-networkd with a simple config.

[Match]
PermanentMACAddress=xxxxx
Name=enp1s0

[Network]
DHCP=no
Address=10.1.10.11/24
DNS=10.1.10.1 84.200.69.80 84.200.70.40
Gateway=10.1.10.1
IPv6AcceptRA=false
LinkLocalAddressing=no

And now it works fine from host to VM, from outside to VM and from different VLAN to VM. For unknown reason (to me) this fails to work with Network Manager but works with Wicked and systemd-networkd.

I don’t understand how NetworkManager can be involved in use of systemd-network. The former should be disabled or purged when using the latter. It’s a replacement.

Netowork Manager is not in use (disabled). Only systemd-networkd is enabled and running.

My last sentence was confusing in previous message. I meant that the network communication did not work properly with Network Manager. The config is purely for systemd-networkd.

If this means the problem is solved, please mark the thread solved for the search engines. :slight_smile:

For some reason the DNS part does not work but requires /etc/resolv.conf file as well with nameservers.

Edit: and this was caused by not running systemd-resolved. A tip for those who might have similar issues.

This was covered in post #2. With systemd-network and static IP especially, resolv.conf needs no “management” from any service like systemd-resolved. /etc/resolv.conf simply needs to exist with suitable content.

I had this as well but when I started systemd-networkd the content vanished (originally created during installation and not touched since). After starting the systemd-resolved, /etc/resolv.conf was properly linked to /run/systemd/network/resolv.conf by systemd-resolved (I assume) which content is based on enp1s0.network configuration. But as mentioned, I did not start systemd-resolved thus the problem with DNS.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.