Bug 2250187

Description of problem:
The startup of nova-compute container starts to fail during the overcloud deployment of typically HCI nodes (nodes with compute delay enabled).
The stdout from the container:
2023-11-15T18:57:04.671122265+00:00 stderr F + echo 'Running command: '\''/var/lib/nova/delay-nova-compute --delay 180 --nova-binary '\'''
2023-11-15T18:57:04.671134462+00:00 stdout F Running command: '/var/lib/nova/delay-nova-compute --delay 180 --nova-binary '
2023-11-15T18:57:04.671148397+00:00 stderr F + umask 0022
2023-11-15T18:57:04.671165844+00:00 stderr F + exec /var/lib/nova/delay-nova-compute --delay 180 --nova-binary
2023-11-15T18:57:04.701564579+00:00 stderr F usage: delay-nova-compute [-h] [--config-file NOVA_CONFIG]
2023-11-15T18:57:04.701564579+00:00 stderr F                           [--nova-binary NOVA_BINARY] [--delay DELAY]
2023-11-15T18:57:04.701564579+00:00 stderr F                           [--state-dir STATE_DIR] [--state-file STATE_FILE]
2023-11-15T18:57:04.701618896+00:00 stderr F delay-nova-compute: error: argument --nova-binary: expected one argument

The compute startup delay is typically enabled for HCI nodes to give ceph some time.
It seems that the problem comes from the recent revert patch:
https://code.engineering.redhat.com/gerrit/plugins/gitiles/openstack-tripleo-heat-templates/+/118d29c35075e0832cb78b5f6b9106cb4e477aa3
Which changed in /deployment/nova/nova-compute-container-puppet.yaml, this:
      kolla_config:
        /var/lib/kolla/config_files/nova_compute.json:
          command:
            list_join:
            - ' '
            - - if:
                - compute_startup_delay
                 - str_replace:
                     template: '/var/lib/nova/delay-nova-compute --delay DELAY --nova-binary'
                     params: { DELAY: {get_attr: [RoleParametersValue, value, nova_compute_startup_delay]} }
-              - if:
-                - {get_param: EnableInstanceHA}
-                - /var/lib/nova/instanceha/check-run-nova-compute
                 - /usr/bin/nova-compute
               - get_attr: [NovaLogging, cmd_extra_args]

to this:

      kolla_config:
        /var/lib/kolla/config_files/nova_compute.json:
          command:
            list_join:
            - ' '
            - - if:
                - compute_startup_delay
                - str_replace:
                    template: '/var/lib/nova/delay-nova-compute --delay DELAY --nova-binary'
                    params: { DELAY: {get_attr: [RoleParametersValue, value, nova_compute_startup_delay]} }
                - /usr/bin/nova-compute
              - get_attr: [NovaLogging, cmd_extra_args]

Previously the possible commands which could be executed (depends on whether compute delay or Instance HA are enabled):
1. /var/lib/nova/delay-nova-compute --delay DELAY --nova-binary /usr/bin/nova-compute
2. /var/lib/nova/delay-nova-compute --delay DELAY --nova-binary /var/lib/nova/instanceha/check-run-nova-compute
3. /usr/bin/nova-compute
4. /var/lib/nova/instanceha/check-run-nova-compute

But now, It's one of the options:
1. /var/lib/nova/delay-nova-compute --delay DELAY --nova-binary
2. /usr/bin/nova-compute

It should be tho I believe:
1. /var/lib/nova/delay-nova-compute --delay DELAY --nova-binary /usr/bin/nova-compute
2. /usr/bin/nova-compute



Version-Release number of selected component (if applicable):
openstack-tripleo-heat-templates-14.3.1-17.1.20231103010821.e7c7ce3.el9ost.noarch

How reproducible:
Always

Steps to Reproduce:
1. Deploy the ovecloud with HCI nodes (nodes with nova-compute delay enabled)

Actual results:
The startup of nove-compute container during the deployment fails.

Expected results:
Successful startup of the container with defined delay and nova-compute service registration.

Additional info: