Machinesets using 4.12 boot images cannot create new nodes. Nodes do no join to the cluster.

    

4.22.0-0.nightly-2026-01-08-022217
    

Always
    

    1. Configure a machineset to use a 4.12 bootimage

    2. Scale up the machineset to create a new node

It can be automatically reproduced by running the automated test case: 
"[sig-mco] MCO scale Author:sregidor-NonHyperShiftHOST-NonPreRelease-Longduration-Critical-76471-[P1] Scaleup using 4.12 cloud image[Disruptive] [Serial]"


    

The node never joins the cluster.

    

The machineset should be able to use the 4.12 boot image to create a new node that joins the cluster.

    

When we access the new machine we can see that there are failing units

[systemd]
Failed Units: 1
  nmstate.service
[core@ip-10-0-31-219 ~]$ journalctl  -u nmstate.service
-- Logs begin at Thu 2026-01-08 11:15:10 UTC, end at Thu 2026-01-08 11:32:23 UTC. --
Jan 08 11:15:56 ip-10-0-31-219 systemd[1]: Starting Apply nmstate on-disk state...
Jan 08 11:15:56 ip-10-0-31-219 systemd[1]: nmstate.service: Main process exited, code=exited, status=203/EXEC
Jan 08 11:15:56 ip-10-0-31-219 systemd[1]: nmstate.service: Failed with result 'exit-code'.
Jan 08 11:15:56 ip-10-0-31-219 systemd[1]: Failed to start Apply nmstate on-disk state.
Jan 08 11:15:56 ip-10-0-31-219 systemd[1]: nmstate.service: Consumed 1ms CPU time


and a failure is reported in the  mco firstboot service
$ journalctl -u  machine-config-daemon-firstboot.service

08 11:33:23 ip-10-0-31-219 podman[2089]: I0108 11:33:23.123930    2113 update.go:1786] Updating files
Jan 08 11:33:23 ip-10-0-31-219 podman[2089]: I0108 11:33:23.123936    2113 command_runner.go:24] Running captured: systemctl list-units --all --no-pager --output=json
Jan 08 11:33:23 ip-10-0-31-219 podman[2089]: I0108 11:33:23.128674    2113 update.go:2819] Removing SIGTERM protection
Jan 08 11:33:23 ip-10-0-31-219 podman[2089]: W0108 11:33:23.128686    2113 firstboot_complete_machineconfig.go:70] error: error parsing systemd units: invalid character 'U' looking for begi>
Jan 08 11:33:23 ip-10-0-31-219 podman[2089]: I0108 11:33:23.128690    2113 firstboot_complete_machineconfig.go:71] Sleeping 1 minute for retry


You can find the must-gather file and journal logs in the comments