[Bug 1770082] Re: systemd-networkd not renaming devices on boot
Ryan Harper
1770082 at bugs.launchpad.net
Fri May 25 14:12:36 UTC 2018
"This is why I added
cloud-init to affected packages -- cloud-init should not be second-
guessing the network layer and attempting to do renames / to run"
There is no second guessing. In the case where we have no network
config, there is no renaming; we accept whatever name is given.
If the config passed to cloud-init includes a name for an interface, then
cloud-init applies that name; this is the MAAS scenario;
Given the unreliable nature of udev w.r.t naming (see this very bug)
The *only* way for cloud-init to ensure that a directive to name an interface
matches the config (also note cloud-init accepts network config in
various formats
not just netplan) is to handle naming if requested directly, precisely
due to this bug.
If we fix systemd-udevd to allow renames of interfaces reliably that helps
most cases where udevd runs. For the remaining cases where udev doesn't run,
containers for example, cloud-init will still need to use iproute2 to
set an interface
name if requested.
On Fri, May 25, 2018 at 8:55 AM, Mathieu Trudel-Lapierre
<mathieu.tl at gmail.com> wrote:
> netplan changes are available in git:
>
> Daniel's patch to write udev rules (SRU material):
> https://github.com/CanonicalLtd/netplan/commit/b0c51bfa8ba8b898a9feaed9cd7d8790d147d35d
>
> Daniel's patch + dropping replug code + rework 'netplan apply' (code for cosmic); in progress for upload to cosmic:
> https://github.com/CanonicalLtd/netplan/tree/live-rename
>
> --
> You received this bug notification because you are subscribed to
> netplan.
> Matching subscriptions: netplan
> https://bugs.launchpad.net/bugs/1770082
>
> Title:
> systemd-networkd not renaming devices on boot
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/netplan/+bug/1770082/+subscriptions
--
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/1770082
Title:
systemd-networkd not renaming devices on boot
Status in netplan:
Confirmed
Status in cloud-init package in Ubuntu:
Confirmed
Status in systemd package in Ubuntu:
Confirmed
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
More information about the foundations-bugs
mailing list