netplan is not doing a daemon-reload afer writing unit configuration on Ubuntu Core

[Alfonso Sanchez-Beato]

When using this cloud.conf file in a gadget snap:

#cloud-config
datasource_list: [None]
users:
  - name: ubuntu
    sudo: "ALL=(ALL) NOPASSWD:ALL"
    lock_passwd: false
    shell: /bin/bash
    passwd: "..."
    ssh_authorized_keys:
      - "..."

On both UC22 and UC24, cloud-init shows this warning in /run/cloud-init/status.json:
...
    "WARNING": [
     "Running ['netplan', 'apply'] resulted in stderr output: Warning: The unit file, source configuration file or drop-ins of netplan-ovs-cleanup.service changed on disk. Run 'systemctl daemon-reload' to reload units.\nWARNING:root:Cannot call Open vSwitch: ovsdb-server.service is not running.\nFailed to connect system bus: Connection refused\nWARNING:root:Falling back to a hard restart of systemd-networkd.service\n"
    ]
...

doing "systemctl cat" for the affected units shows a similar error message.

There is a difference on behavior between UC22 and UC24 as in the former cloud-init finishes without error, but in the latter it finishes with a "recoverable error" code due to the netplan issue. However the same warning is shown.

[Lukas Märdian (slyon)]

Do we know the Netplan versions used in those UC22 & UC24 images?

This might also be related to https://github.com/canonical/netplan/pull/421 which should be fixed as of Netplan v1.0 in Noble (and UC24?).

[Alfonso Sanchez-Beato (alfonsosanchezbeato)]

cloud-init=24.1.3-0ubuntu3 in UC24
cloud-init=24.1.3-0ubuntu1~22.04.1 in UC22

In general versions in base snaps should be the same as in classic, maybe with a bit of a delay on updates.

[Brett Holman (holmanb)]

To understand order of events with cloud-init, please run `cloud-init collect-logs` and attach the generated tarball.

[Alfonso Sanchez-Beato (alfonsosanchezbeato)]

Tarball generated by "sudo cloud-init collect-logs"