[Bug 1929560] Re: link mac isn't set if already up

[Jan Graichen]

Thanks a lot for your work, Dan! Unfortunately, I wasn't able to pick up
the new bug from here, sorry for that. (It appears I did not get any
notification on your comments, need to check that)

The bridges are managed by OpenStack Neutron from Ubuntu cloud archives
and with names derived from the network UUID in the cluster database. It
might work for a very limited set of bridges, if they do not set
different options. AFAIK, OpenStack Neutron is still working on
supporting using external bridges
(https://bugs.launchpad.net/neutron/+bug/1105488).

As we have bridges on VLANs on bonded interfaces (currently managed with
systemd-networkd), I will see if I have to replace that all or if there
is an option to just have systemd-networkd not set nomaster...

Thanks!

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to systemd in Ubuntu.
https://bugs.launchpad.net/bugs/1929560

Title:
  link mac isn't set if already up

Status in systemd:
  Fix Released
Status in systemd package in Ubuntu:
  Fix Released
Status in systemd source package in Bionic:
  Fix Released
Status in systemd source package in Focal:
  Fix Released
Status in systemd source package in Groovy:
  Fix Released

Bug description:
  [impact]

  if link is already up, configured mac isn't set

  this also affects the master/nomaster configuration

  [test case]

  create .network config for the link, for example:

  [Match]
  Name=eth0

  [Link]
  MACAddress=00:16:3e:ff:ff:ff

  [Network]
  DHCP=yes

  
  make sure the link is already up, and restart systemd-networkd:

  root at test-f:/etc/systemd/network# ip l show eth0
  61: eth0 at if62: <BROADCAST,MULTICAST> mtu 1500 qdisc noqueue state DOWN mode DEFAULT group default qlen 1000
      link/ether 00:16:3e:fc:b2:00 brd ff:ff:ff:ff:ff:ff link-netnsid 0
  root at test-f:/etc/systemd/network# ip l set dev eth0 up
  root at test-f:/etc/systemd/network# ip l show eth0
  61: eth0 at if62: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
      link/ether 00:16:3e:fc:b2:00 brd ff:ff:ff:ff:ff:ff link-netnsid 0
  root at test-f:/etc/systemd/network# systemctl restart systemd-networkd
  root at test-f:/etc/systemd/network# ip l show eth0
  61: eth0 at if62: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
      link/ether 00:16:3e:fc:b2:00 brd ff:ff:ff:ff:ff:ff link-netnsid 0
  root at test-f:/etc/systemd/network# ip l set dev eth0 down
  root at test-f:/etc/systemd/network# ip l show eth0
  61: eth0 at if62: <BROADCAST,MULTICAST> mtu 1500 qdisc noqueue state DOWN mode DEFAULT group default qlen 1000
      link/ether 00:16:3e:fc:b2:00 brd ff:ff:ff:ff:ff:ff link-netnsid 0
  root at test-f:/etc/systemd/network# systemctl restart systemd-networkd
  root at test-f:/etc/systemd/network# ip l show eth0
  61: eth0 at if62: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
      link/ether 00:16:3e:ff:ff:ff brd ff:ff:ff:ff:ff:ff link-netnsid 0

  [regression potential]

  any regression would likely result in incorrectly set link mac and/or
  master, or failure to correctly set mac and/or master

  [scope]

  this is needed for g and earlier

  this is fixed upstream by commit
  e16e4b3bedc57919b3ecb4096e941709a5d39b8a which is included in v247 so
  this is fixed already in h and later

  [other info]

  this moves the configuration of the link mac and master out of the
  link_up() function, so this also allows repeated calls to link_up()
  without reconfiguring the link mac and master. That behavior is useful
  for the commits for bug 1664844

To manage notifications about this bug go to:
https://bugs.launchpad.net/systemd/+bug/1929560/+subscriptions