NetworkManager ignores my config
This is a laptop running Fedora 28. Today NetworkManager started to ignore my configuration settings, which had worked for years, literally.
I want:
- physical interface enp0s25 plugged into bridge br1
- no IP address on enp0s25
- br1 gets IP address via DHCP
Since today, NM doesn't plug enp0s25 into the bridge. It runs DHCP on enp0s25 and sets an IP address. Since br1 is not connected to the outside world, its DHCP fails.
I use this laptop as a VM host. When I start a VM, libvirt plugs its interface (named vnet0 etc) into br1. However now, a few seconds later, NM removes the interface from br1. It didn't behave like this yesterday.
To add insult to injury, NM runs even after I disable it and reboot:
$ systemctl status NetworkManager
● NetworkManager.service - Network Manager
Loaded: loaded (/usr/lib/systemd/system/NetworkManager.service; disabled; vendor preset: enabled)
Active: active (running) since Thu 2018-06-21 20:18:22 JST; 2min 23s ago
The journal just reports what NM is doing. It doesn't say that my config is invalid, nor does it give any other reason why it is ignored.
How can I find out where the problem is?
The interface and the bridge are configured as follows. Adding NM_CONTROLLED=no leaves them unconfigured; I can then ifup them manually and attach my VMs to the network as before:
$ cat ifcfg-enp0s25
TYPE=Ethernet
BOOTPROTO=none
NAME=enp0s25
DEVICE=enp0s25
ONBOOT=yes
BRIDGE=br1
$
$ cat ifcfg-br1
TYPE=Bridge
BOOTPROTO=dhcp
DEFROUTE=yes
PEERDNS=yes
PEERROUTES=yes
NAME=br1
DEVICE=br1
ONBOOT=yes
And this is the situation that results from this configuration:
$ brctl show
bridge name bridge id STP enabled interfaces
br1 8000.000000000000 no
$
$ ip a show dev enp0s25
2: enp0s25: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
link/ether f0:de:f1:62:b9:f1 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.24/24 brd 192.168.1.255 scope global dynamic noprefixroute enp0s25
valid_lft 7088sec preferred_lft 7088sec
inet6 fe80::f2de:f1ff:fe62:b9f1/64 scope link noprefixroute
valid_lft forever preferred_lft forever
$
$ ip a show dev br1
4: br1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default qlen 1000
link/ether de:db:4a:cd:8b:96 brd ff:ff:ff:ff:ff:ff
Did you upgrade NetworkManager yesterday? It might be a regression that needs to be reported to bugzilla
Thanks. I upgraded on Tuesday, but the symptoms appeared on Thursday after I used this laptop to install DD-WRT on a router, which required fiddling with LAN and WLAN. I tend to think that I fiddled too much.
What NM settings can I check outside of the config files?
From my command history (no, I didn't know what I was doing. Yes, I know I shouldn't have):