Right. Yes, I can replicate that, thanks heaps! So to summarise, you can make the rename take effect on boot if you: 1) copy the files from /run/systemd/network -> /etc/systemd/network 2) then update-initramfs -u
This seems pretty far outside of the way that netplan is supposed to work - indeed using /etc/systemd/ is basically bypassing netplan. So we still have the issue that just changing 'set-name' in /etc/netplan/*.yaml doesn't work as expected. What should we change so that set-name works as expected? I see a few options: 0) Document that set-name is fragile and stop relying on it. This is actually really easy to do and I have a netplan patch drafted already. The reason that we have no network connectivity when set-name fails is that the network file netplan creates maches on *both* the new name and the mac address. We can just drop the new name from the [Match] section of the relevant .network file and match only on the provided mac address (or whatever else was used in the netplan match stanza). This means that regardless of the interface name it's brought up and networking works. 1) Make the initramfs hook for udev copy the files from /run (as well as from /lib and /etc) into the initial ramdisk. This is probably my preference; we can run netplan generate beforehand to make sure we get a clean copy, and then document that update-initramfs is required to get stuff to stick. 2) Allow udev to re-rename interfaces. I don't know if this would be acceptable to systemd upstream? 3) Something else? Regards, Daniel -- You received this bug notification because you are a member of Ubuntu Touch seeded packages, which is subscribed to systemd in Ubuntu. https://bugs.launchpad.net/bugs/1770082 Title: systemd-networkd not renaming devices on boot Status in netplan: Incomplete Status in systemd package in Ubuntu: New Bug description: === 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. To manage notifications about this bug go to: https://bugs.launchpad.net/netplan/+bug/1770082/+subscriptions -- Mailing list: https://launchpad.net/~touch-packages Post to : touch-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~touch-packages More help : https://help.launchpad.net/ListHelp
