Description of problem:

During FFU, and After running the the overcloud upgrade run, the ovn_controller container fails to start with the below error:

~~~
"ERROR: Container ovn_controller exited with code 125 when runed\nstderr: Error: invalid value all for cpuset cpus\n"]}
~~~

it seems it comes from here:
$ cat hashed-ovn_controller.json
{
"cpuset_cpus": "all", <======
"depends_on": [
"openvswitch.service"
],
"environment": {
"KOLLA_CONFIG_STRATEGY": "COPY_ALWAYS"

Checking the Cu templates, The parameter OVNContainerCpusetCpus is defined for all Compute roles (various roles), but not defined for controllers nor Networker nodes (where the issue is happening).

To overcome the issue (pass the error), the parameter cpuset_cpus:0 was set 'manually' to '0' to all issued nodes on the hashed ovn_controller file., and then overcloud upgrade run

• Cu needs to know where the cpuset_cpus": "all" came from ?
• How to fix this issue without manually change this parameter or through templates ?
• Why did the error pass and then ovn_controller become up after defining this parameter to cpuset_cpus": "0" ?.

Version-Release number of selected component (if applicable):
openstack-ovn-controller:16.2.6

How reproducible:
NA

Steps to Reproduce:
1.
2.
3.

Actual results:
After running the the overcloud upgrade run, the ovn_controller container fails to start

Expected results:
the ovn_controller become up with the default cpuset_cpus value After running the the overcloud upgrade run step