Is there a document somewhere that shows what the systemd team thinks is a functional boundary? What things they don't plan on replacing?
Anyway, I reported it and it presumably got fixed. I wouldn't know, because systemd dev handling this didn't say a thing to me at all. Just closed the report after some seemingly not very related commits to what I saw in gdb.
I just dropped the pam module from config anyway, and I'm not willing to lose access to my cluster again just to try.
Not sure how distro fucked up in this case, when it's homed pam module going into infinite loop once in a while unpredictably blocking all processes that use pam naively, unless you mean that they fucked up by using systemd or pam at all.
Edit: lol, I read through systemd release notes and there's no mention of new huge systemd-homed pam module at all. But good I looked, they're moving pam config for systemd to /usr/lib/pam.d and systemd-home pam module is again enabled there. So I'll have to check again all my systems if it's really disabled after upgrade.