Re: FC31 : won't activate interfaces - "strictly unmanaged"

[sean darcy <seandarcy2 gmail com>]

On 1/26/20 9:17 PM, sean darcy via networkmanager-list wrote:
> On 1/26/20 3:16 AM, Thomas Haller via networkmanager-list wrote:
> > On Sat, 2020-01-25 at 14:34 -0500, sean darcy via networkmanager-list
> > wrote:
> > > Fedora FC31. NetworkManager-1.20.10-1.fc31.x86_64
> > > This a cross post with Fedora user.
> > >
> > > On FC31 I can't persistently rename the interfaces, see:
> > > https://bugzilla.redhat.com/show_bug.cgi?id=1777893
> > >
> > > I'll give up and use FC31 interface names : enp1s0 and enp0s20u3.
> > >
> > > I've used nmtui to set up the interfaces. But NM won't activate them
> > > because they are "strictly unmanaged" !!
> >
> > Hi,
> >
> >
> > I think this is a bug.
> >
> > NetworkManager is running in initrd. Thereby it already activates the
> > interface. Later, udev wants to rename the interface, but since it's
> > already up, it fails to do so.
> >
> > The result is that udev doesn't indicate that the device is
> > initialized, which causes NM to wait for it (and thus it's unmanaged).
> >
> > It seems there are several potential issues:
> >
> >    - why does NetworkManager even run in initrd, if you didn't configure
> > it to do so?
> >
> >    - if NM is running in initrd, should it wait for udev to rename the
> > device?
> >
> >    - why does udev completely fail in this case?
> >
> >    - maybe NM should after a timeout stop waiting for udev, to mitigate
> > this issue.
> >
> >
> > You could look at the logfile to verify that. Enable level=TRACE
> > logging, but most interesting are also the messages from initrd and
> > udev.
> >
> >
> > best,
> > Thomas
> >
> > > What does that mean and how do I fix it ?
> > >
> > > Is there some other way to persistently activate the interface ?
> > >
> > > /etc/network/interfaces does not exist.
> > >
> > > sean
> > >
> > >    cat /etc/sysconfig/network-scripts/ifcfg-enp*
> > > TYPE=Ethernet
> > > PROXY_METHOD=none
> > > BROWSER_ONLY=no
> > > BOOTPROTO=none
> > > IPADDR=10.10.11.251
> > > PREFIX=24
> > > DNS1=10.10.11.251
> > > DEFROUTE=no
> > > IPV4_FAILURE_FATAL=no
> > > IPV6INIT=no
> > > IPV6_AUTOCONF=yes
> > > IPV6_DEFROUTE=yes
> > > IPV6_FAILURE_FATAL=no
> > > IPV6_ADDR_GEN_MODE=stable-privacy
> > > NAME=enp0s20u3
> > > UUID=bf8aa45c-31e4-403a-9c14-c1a59bf0f2b2
> > > ONBOOT=yes
> > > DEVICE=enp0s20u3
> > >
> > > TYPE=Ethernet
> > > PROXY_METHOD=none
> > > BROWSER_ONLY=no
> > > BOOTPROTO=dhcp
> > > DEFROUTE=yes
> > > IPV4_FAILURE_FATAL=no
> > > IPV6INIT=yes
> > > IPV6_AUTOCONF=yes
> > > IPV6_DEFROUTE=yes
> > > IPV6_FAILURE_FATAL=no
> > > IPV6_ADDR_GEN_MODE=stable-privacy
> > > NAME=enp1s0
> > > UUID=a1b089e9-5a13-4df7-8f09-9f54ebc56cea
> > > DEVICE=enp1s0
> > > ONBOOT=yes
>
> Thanks for the quick response. I hope it's not a bug, since I'd like to 
> get this machine up and running.
>
> "Enable level=TRACE" where ? rsyslogd.conf ? For kernel ? udevd? 
> everything ?
>
> sean

I put net.ifames=0 in the kernel command line. I removed the 
/etc/udev/rules/70-persistent-net.rules. I set up 
/etc/sysconfig/network-scripts/ifcfg-eth[01].

On reboot, I had eth0 and eth1. eth1 was active. But eth0 was not 
active. When I used nmtui to activate eth0. I got the dreaded "strictly 
unmanaged" error.

Here's dmesg | grep eth0

:25 lenovo-gateway.riverside kernel: r8169 0000:01:00.0 eth0: 
RTL8168gu/8111gu, 50:7b:9d:0b:8a:ab, XID 509, IRQ 44
:25 lenovo-gateway.riverside kernel: r8169 0000:01:00.0 eth0: jumbo 
features [frames: 9200 bytes, tx checksumming: ko]
:27 lenovo-gateway.riverside NetworkManager[484]: <info> 
[1580149767.5593] manager: (eth0): new Ethernet device 
(/org/freedesktop/NetworkManager/Devices/2)
:27 lenovo-gateway.riverside NetworkManager[484]: <info> 
[1580149767.5594] device (eth0): state change: unmanaged -> unavailable 
(reason 'managed', sys-iface-state: 'external')
:27 lenovo-gateway.riverside kernel: r8169 0000:01:00.0 eth0: Link is Down
:30 lenovo-gateway.riverside kernel: r8169 0000:01:00.0 eth0: Link is Up 
- 1Gbps/Full - flow control off
:30 lenovo-gateway.riverside kernel: IPv6: ADDRCONF(NETDEV_CHANGE): 
eth0: link becomes ready
:30 lenovo-gateway.riverside NetworkManager[484]: <info> 
[1580149770.5499] device (eth0): carrier: link connected
:30 lenovo-gateway.riverside NetworkManager[484]: <info> 
[1580149770.5502] device (eth0): state change: unavailable -> 
disconnected (reason 'carrier-changed', sys-iface-state: 'managed')
:55 lenovo-gateway.riverside systemd-udevd[695]: eth0: Failed to rename 
network interface 2 from 'eth0' to 'external': Device or resource busy
:55 lenovo-gateway.riverside systemd-udevd[695]: eth0: Failed to process 
device, ignoring: Device or resource busy
Jan 27 13:30:33 lenovo-gateway.riverside NetworkManager[975]: <info> 
[1580149833.3056] device (eth0): carrier: link connected
Jan 27 13:30:33 lenovo-gateway.riverside NetworkManager[975]: <info> 
[1580149833.3059] manager: (eth0): new Ethernet device 
(/org/freedesktop/NetworkManager/Devices/4)


Note at :55 udev tries to rename the device. But there's NO udev rule to 
do so. So where's that from ??


grep -R external /etc/udev/rules.d/*
/etc/udev/rules.d/40-libsane.rules:# If your scanner is supported by 
some external backend (brother, epkowa,
/etc/udev/rules.d/S99-2000S1.rules:# If your scanner is supported by 
some external backend (brother, epkowa,

It seems udev tries to rename the device, and fails because the 
"resource is busy", whatever that means. But then NetworkManager won't 
activate it.

I'm surprised this problem hasn't arisen with every multihomed machine, 
but there are no posts I've found.

How do I get udev not to try to rename eth0 ? But that's a udev problem.

This is the NetworkManager list. How do I force NetworkManager to 
"manage" this interface ?

sean