systemd-tmpfiles doesn't create /run/postgresql at reboot, but does when called manually with --create

I have a problem related to a few already answered here in the past, but the solutions offered for them already appear to be in place for me: - [/var/run/postgresql missing after reboot](https://serverfault.com/questions/766287/) - [Tired of creating /run/postgresql and setting read and execute writes after every reboot](https://stackoverflow.com/questions/41515640/) - [Postgresql failed to start. Reboot is resetting the permission given to /var/run/postgresql/](https://stackoverflow.com/questions/58850796/) For me, I have postgresql running on my cluster(s), but it doesn't automatically start back up when the cluster is rebooted (which happens periodically for automated security patching). The problem is that /run/postgresql isn't created after reboot. However, the config to do so appears to already be in place and correct:

# cat /usr/lib/tmpfiles.d/postgresql-16.conf
d /run/postgresql 0755 postgres postgres -

I've tried rebooting a cluster and verified that the directory isn't created - but then if I run systemd-tmpfiles --create manually, it *does* get created and I'm able to start postgresql and proceed as normal. Everything else in /run seems to be getting created correctly - it's specifically postgresql that has a problem. Looking at the systemd-tmpfiles status:

# service  systemd-tmpfiles-setup status
Redirecting to /bin/systemctl status systemd-tmpfiles-setup.service
● systemd-tmpfiles-setup.service - Create Volatile Files and Directories
...
Jun 16 05:59:21 ...hostname... systemd: Starting Create Volatile Files and Directories...
Jun 16 05:59:21 ...hostname... systemd-tmpfiles: [/usr/lib/tmpfiles.d/httpd.conf:1] Unknown group 'apache'.
Jun 16 05:59:21 ...hostname... systemd-tmpfiles: [/usr/lib/tmpfiles.d/httpd.conf:2] Unknown user 'apache'.
Jun 16 05:59:21 ...hostname... systemd-tmpfiles: [/usr/lib/tmpfiles.d/postgresql-16.conf:1] Unknown user 'postgres'.
Jun 16 05:59:21 ...hostname... systemd: Started Create Volatile Files and Directories.

So it looks like the issue is that some users (coming from a Red Hat Identity Management server) aren't available at the time the /run dir is initially created. I'm not sure how to manipulate the timing for this... I'm using RHEL 8.10 and postgres16 - but I've seen similar issues with RHEL9 and postgres12 as well. Everything is coming from yum so the packages, configs, etc. should be "vanilla".