[Bug 1770082] Re: systemd-networkd not renaming devices on boot

Thu Jul 5 13:31:49 UTC 2018

** Description changed:

+ [Impact]
+ Systems relying on renaming network interfaces at boot and when 'netplan apply' is run.
+ 
+ [Test case]
+ - Write a new netplan YAML (adjusting for current system as necessary):
+ network:
+     version: 2
+     ethernets:
+         ens3:
+             dhcp4: true
+             match:
+                 macaddress: 52:54:00:de:bd:f6
+             set-name: myif0
+ 
+ - Run 'netplan apply'
+ - Verify that the device is correctly renamed to 'myif0'.
+ - Reboot.
+ - Make sure the device is correctly renamed to 'myif0'.
+ 
+ [Regression potential]
+ Changes in rename logic to add udev rules may otherwise impact applying different settings to the network interfaces. Changes in settings on network interfaces, missing parameters (especially on bonds, bridges) should be investigated as potential regressions. Other failures to apply network settings might also happen if there's a race between applying renames via the udev rules, and using the new names to apply configuration changes to the interfaces.
+ 
  === systemd issue ===

  Renaming devices doesn't seem to work.

  If I disable all other network configuration and create
  /etc/systemd/network/10-network.link with:

  [Match]
  MACAddress=52:54:00:c1:c9:bb

  [Link]
  Name=myiface3

  I expect this to cause the device with that MAC address to be renamed to
  myiface3. However, when I reboot, I instead see:

  $ ip l
  1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
      link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
  2: ens3: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
      link/ether 52:54:00:c1:c9:bb brd ff:ff:ff:ff:ff:ff

  The device is not renamed.

  This link file is pretty much identical to Example 2 in
  https://www.freedesktop.org/software/systemd/man/systemd.link.html.

  The renaming does work if I boot with net.ifnames=0, and oddly, it also
  works if I unbind the device and rebind it as netplan apply does. No
  setting of NamePolicy seems to help.

  === Original Bug ==

  'set-name:' doesn't change the name of a network interface on boot, it
  only works when you do netplan apply.

  Say I take this 50-cloud-init.yaml file:

  # This file is generated from information provided by
  # the datasource.  Changes to it will not persist across an instance.
  # To disable cloud-init's network configuration capabilities, write a file
  # /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg with the following:
  # network: {config: disabled}
  network:
      version: 2
      ethernets:
          ens3:
              dhcp4: true
              match:
                  macaddress: 52:54:00:de:bd:f6
              set-name: ens3

  Say I change set-name to 'myiface3' and reboot. I expect that the device
  will be called myiface3 and brought up fine with dhcp. However, instead
  I see:

  $ ip a
  1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
      link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
      inet 127.0.0.1/8 scope host lo
         valid_lft forever preferred_lft forever
      inet6 ::1/128 scope host
         valid_lft forever preferred_lft forever
  2: ens3: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
      link/ether 52:54:00:de:bd:f6 brd ff:ff:ff:ff:ff:ff

  The name has not been changed, and the device has not been brought up.

  If I run netplan apply however, I see the following:

  1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
      link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
      inet 127.0.0.1/8 scope host lo
         valid_lft forever preferred_lft forever
      inet6 ::1/128 scope host
         valid_lft forever preferred_lft forever
  3: myiface3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
      link/ether 52:54:00:de:bd:f6 brd ff:ff:ff:ff:ff:ff
      inet 192.168.122.151/24 brd 192.168.122.255 scope global dynamic myiface3
         valid_lft 3575sec preferred_lft 3575sec
      inet6 fe80::5054:ff:fede:bdf6/64 scope link
         valid_lft forever preferred_lft forever

  So names are successfully changed with netplan apply.

  This seems to be some udev-related timing or priority issue that I'm
  still trying to hunt down.

  This breaks some forms of migration in certain cloud environments.

-- 
You received this bug notification because you are a member of Ubuntu
Sponsors Team, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1770082

Title:
  systemd-networkd not renaming devices on boot

Status in netplan:
  Confirmed
Status in cloud-init package in Ubuntu:
  Confirmed
Status in netplan.io package in Ubuntu:
  Fix Released
Status in nplan package in Ubuntu:
  Fix Released
Status in systemd package in Ubuntu:
  Confirmed

Bug description:
  [Impact]
  Systems relying on renaming network interfaces at boot and when 'netplan apply' is run.

  [Test case]
  - Write a new netplan YAML (adjusting for current system as necessary):
  network:
      version: 2
      ethernets:
          ens3:
              dhcp4: true
              match:
                  macaddress: 52:54:00:de:bd:f6
              set-name: myif0

  - Run 'netplan apply'
  - Verify that the device is correctly renamed to 'myif0'.
  - Reboot.
  - Make sure the device is correctly renamed to 'myif0'.

  [Regression potential]
  Changes in rename logic to add udev rules may otherwise impact applying different settings to the network interfaces. Changes in settings on network interfaces, missing parameters (especially on bonds, bridges) should be investigated as potential regressions. Other failures to apply network settings might also happen if there's a race between applying renames via the udev rules, and using the new names to apply configuration changes to the interfaces.

  === systemd issue ===