degraded network on veth*

[Pan Luo]

Hi,

Is it normal to see "degraded" on the network status of the veth* interfaces? I saw it on our production servers (VMware) and the fresh new vagrant image.

Steps to replicate:

git clone https://github.com/coreos/coreos-vagrant.git
cd coreos-vagrant
mv user-data.example user-data
mv config.rb.sample config.rb
vagrant up
docker run --rm -it alpine sh

# open another terminal and vagrant ssh into VM
networkctl status veth27a7977

● 9: veth27a7977
       Link File: /usr/lib64/systemd/network/99-default.link
    Network File: n/a
            Type: ether
           State: degraded (unmanaged)
          Driver: veth
      HW Address: 06:d0:3d:64:7c:c1
         Address: fe80::4d0:3dff:fe64:7cc1

As you can see , the network interface status is "degraded". Is this normal? Thanks.

Cheers,

Pan

[David Michael]

On Thu, Apr 27, 2017 at 4:33 PM, Pan Luo <luop...@gmail.com> wrote:
> Hi,
>
> Is it normal to see "degraded" on the network status of the veth* interfaces?

Yes, Docker's veths are not managed by networkd, and networkd defines
"degraded" state as "only link or site local addresses found". Is
there a problem with these interfaces?

Thanks.

David

[Pan Luo]

I was having issue about the intermit network connection issues in containers. I found the work around in this issue: https://github.com/coreos/bugs/issues/1802. However, during the troubleshooting process, I saw this: https://github.com/coreos/bugs/issues/1785#issuecomment-279281914. "dm0-" suggests that "90-docker-veth.network" should be in "Network File", but as you can see, mine is n/a. I was wondering if that was the reason the state is "degraded".

[David Michael]

On Thu, Apr 27, 2017 at 6:18 PM, Pan Luo <luop...@gmail.com> wrote:
> I was having issue about the intermit network connection issues in
> containers. I found the work around in this issue:
> https://github.com/coreos/bugs/issues/1802. However, during the
> troubleshooting process, I saw this:
> https://github.com/coreos/bugs/issues/1785#issuecomment-279281914. "dm0-"
> suggests that "90-docker-veth.network" should be in "Network File", but as
> you can see, mine is n/a. I was wondering if that was the reason the state
> is "degraded".

I haven't looked for the cause yet, but I imagine networkd just
reworked what data is saved for different link states. The systemd
version in bug 1785 was 231, while now all channels are using systemd
233. You should still be able to verify correct interface settings
are applied with "ip link" commands.

Thanks.

David

[Pan Luo]

Haha, just realized that dm0- is YOU! Thanks for your help!

Tried "ip link" and it seems it's in the correct state:

6: vethfbf5482@if5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master docker0 state UP mode DEFAULT group default

    link/ether 1e:f5:1e:bc:ae:cb brd ff:ff:ff:ff:ff:ff

8: veth8aa367e@if7: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master docker0 state UP mode DEFAULT group default

    link/ether 42:e7:d0:49:8f:5d brd ff:ff:ff:ff:ff:ff

--

Cheers,

Pan