A restart of slapd on serpens was requested in response to certificate rotation on 2020-12-03 14:20:12.
Dec 03 14:20:42 serpens systemd[1]: slapd.service: Found left-over process 6300 (slapd) in control group while starting unit. Ignoring.
The process hung around for 7 hours following the SIGTERM and remained stuck until ops responded to alerting. The remedy was to issue a SIGKILL and restart slapd manually.
This behavior and remedy seems familiar to the wikitech outage caused by LDAP on seaborgium a few weeks ago.