While investigating https://issues.redhat.com/browse/RHEL-131342?focusedId=28978107&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-28978107

I discovered that the cpus are not being isolated nor other realtime params are being set.

[root@realtime-03 ~]# cat /proc/cmdline
BOOT_IMAGE=(hd0,gpt2)/boot/ostree/default-98706d3687721de269c2274df931c7570c314d0b94281a5ded314094a36303da/vmlinuz-6.12.0-191.el10.x86_64+rt root=/dev/mapper/rhel_realtime--03-root rd.lvm.lv=rhel_realtime-03/root rd.lvm.lv=rhel_realtime-03/swap rw boot=UUID=943e47a7-be07-40cb-91c3-a1e6e765a06b ostree=/ostree/boot.0/default/98706d3687721de269c2274df931c7570c314d0b94281a5ded314094a36303da/0


 cat "/sys/devices/system/cpu/isolated"
# no output 

On a machine freshly converted to Image Mode I was able to set tuned-adm-profile to realtime without a reported issue from the tool, but I do not believe it was being set. On the same server having being switched images (not reprovisioned by beaker, but upgraded or switched out several times) tuned-adm stalls upon setting the realtime profile again (after setting it to throughput-performance from realtime).

What were you trying to do that didn't work?
Set the realtime profile in Image Mode

What is the impact of this issue to you?

Realtime is not set on Image Mode or tuned-adm-profile stalls while setting realtime.

Output after pressing ctrl-c after it stalls:

[root@realtime-03 ~]# sudo tuned-adm profile realtime
^CTraceback (most recent call last):
  File "/sbin/tuned-adm", line 134, in <module>
    result = admin.action(action_name, **options)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/site-packages/tuned/admin/admin.py", line 88, in action
    res = self._controller.run()
          ^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/site-packages/tuned/admin/dbus_controller.py", line 68, in run
    self._main_loop.run()
  File "/usr/lib/python3.12/site-packages/gi/overrides/GLib.py", line 495, in run
    with register_sigint_fallback(self.quit):
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib64/python3.12/contextlib.py", line 144, in __exit__
    next(self.gen)
  File "/usr/lib64/python3.12/site-packages/gi/_ossighelper.py", line 237, in register_sigint_fallback
    signal.default_int_handler(signal.SIGINT, None)
KeyboardInterrupt  

Please provide the package NVR for which the bug is seen:

tuned-2.26.0-1.el10_1.1

How reproducible is this bug?:

I am not sure at this point. This was on a system that had been switched/upgraded with bootc several times. You may need to restart the Image Mode system after setting realtime, toggle the system back to throughput-performance and then back to realtime. Or, use bootc-switch/upgrade <image-name> --apply.

Steps to reproduce

1. Set the system to image mode, I will provide links to a tutorial and Containerfile in the comments. Note this Containerfile will set the system to realtime and reboot one time on first boot.
2. Check the above commands to see if the cpus are isolated or other rt settings are set
3. Try switching/upgrading the image. Check if rt settings are set.
4. Use tuned-adm profile to switch the system back to throughput-performance and then back to realtime. Check if rt settings are set.

Expected results

Linux cpus to be isolated and other rt params to be set.

Actual results

See the very top of this issue.

Only was tested on x86_64 and 10.2.