To Reproduce Steps to reproduce the behavior:
A customer transitioned one compute node, executed upgrade prepare command and tried to upgrade transitioned compute. Upgrade failed because Nova puppets didn't create virtsecretd configuration file and nova_virtsecretd container was unable to start:

2025-05-12T15:19:15.663737992+00:00 stderr F 2025-05-12 15:19:15.663+0000: 211475: error : main:918 : Can't load config file: Failed to open file '/etc/libvirt/virtsecretd.conf': No such file or directory: /etc/libvirt/virtsecretd.conf

I was unable to identify clear pointers explaining this situation: we have templates and overcloud-deploy, but services are set consistently there. At the same time (from hieradata and from some executions) it looks like legacy libvirt service definitions were used in some occasions, while from puppet tags it looks like normal libvirt service definition was used. Please help us isolate this problem and figure out a workaround.

Expected behavior
Puppets generate consistent configuration for libvirt services during upgrade

Bug impact
FFU is blocked

Known workaround
None

Additional context
To be provided privately