Running CentOS7, with openLDAP-2.4.40-13.el7. The environment consists of two ldap providers, in mirror mode, serving over a shared virtual IP. Client-facing services are provided by 4 consumers, most of which are accessed over a layer 4 load balancer.

Periodically, the consumers encounter some sort of client request(s)
which consume all available threads, cause backload threads to spike, and cause slapd to go unresponsive for a long period of time. I’ve no idea what is causing these events, or if there is anything in the configurations that I can tweak to help.

Anyone have any ideas?