Network Manager is generating wrong profiles

[Karl Weber]

Hi,

I observe a strange behaviour of Network Manager and would like to ask, if 
this is a bug or a feature. If it is a feature, I would greatly appreciate an 
explanation about why Network Manager (NW) works the way it does.

On my PC I have two NICs, eth0 and eth1. eth0 is connected to a router and 
configured by the router using DHCP. eth1 is connected to the intranet and 
configured manually.

For both I have a persistent configuration in files /etc/NetworkManager/system-
connections/ethN.nmconnection for N = 0,1. They are given at the end. 
Furthermore, I have a config file /etc/NetworkManager/conf.d/99-serverlike.conf.

My PC runs openSUSE Leap 15.6 with NM out of the BOX, except for the three 
files mentioned.

When I boot this PC, both NICs are configured as expected and working well. The 
same is true, when I restart NM with systemctl restart NetworkManager.

However, when the PC is suspended and waking up again, thing work differently. 

At the beginning, when I did not have the file 99-serverlike.conf at all, NW 
was generating a new profile for eth1 and using this one. With this new 
configuration eth1 was not working at all. The configuration in 
eth1.nmconnection was ignored. eth0, on the other hand, was configured as 
defined through eth0.nmconnection -- and working.

Then, I created the file 99-serverlike.conf and added the [main] section (but 
not yet the [device] section). In this case, NM still created the wrong 
profile, when waking up, but activated the expected profile given in 
eth1.nmconnection. The connection was working, however, the system tray icon 
was "rotating" the whole time, which was very annoying. Until I deleted the 
wrong profile with nmcli or restarted the NM with systemctl.

When the PC woke up and the wrong profile was created, the state transition 
from unavailable to disconnected had:

- reason none and sys-iface-state managed for eth0

- reason connection-assumed and sys-iface-state external for eth1.

Then  I added the [device] section to the 99-serverlike.conf file and the 
problem disappeared. I.e. no wrong profile was created any more and the state 
transition from unavailable to disconnected when waking up also had reason 
none and sys-iface-state managed for eth1.

What is going on here? Why does NM behave so differently for eth0 and eth1?

Thanks in advance for any explanation!

Best regards
Karl

eth0.nmconnection:
=====================

[connection]
id=eth0
uuid=7ba00b1d-8cdd-30da-91ad-bb83ed4f7474
type=ethernet
timestamp=1749982652
zone=external

[ethernet]
mac-address=50:3E:AA:08:34:A9

[ipv4]
method=auto

[ipv6]
addr-gen-mode=stable-privacy
method=auto

[proxy]

eth1.nmconnection:
=====================

[connection]
id=eth1
uuid=5f258f8a-dc4d-4557-9a7c-6f5f8f88f917
type=ethernet
timestamp=1749806262
zone=public

[ethernet]
mac-address=00:D8:61:2E:33:09

[ipv4]
address1=192.168.5.1/24
dns-search=home.arpa;
method=manual

[ipv6]
addr-gen-mode=stable-privacy
address1=fdfd:96f1:f163:bb1d::1/64
dns=fdfd:96f1:f163:bb1d::1;
dns-priority=2
dns-search=home.arpa;
method=shared

[proxy]


99-serverlike.conf
=====================

[main]
no-auto-default=*
ignore-carrier=*

[device]
match-device=interface-name:eth1
keep-configuration=no