High memory consumption with Red Hat directory server 9.1

Issue

After upgrading to RHDS 9.1, memory consumption of the ns-slapd process is continuously growing from a few MB up to approximately 15GB. This behavior is completely different from what we have seen on the same servers before with 8.2 (with the same performance settings like cache or limits) and what we see on other servers where we still use version 8.2.

The ns-slapd process has been killed by OOM killer:

Sep 20 06:15:07 host1 kernel: Out of memory: Kill process 17796 (ns-slapd) score 974 or sacrifice child
Sep 20 06:15:07 host1 kernel: Killed process 17796, UID 2040, (ns-slapd) total-vm:26126472kB, anon-rss:15736732kB, file-rss:1852kB

Sep 20 07:20:49 host1 kernel: Out of memory: Kill process 7376 (ns-slapd) score 958 or sacrifice child
Sep 20 07:20:49 host2 kernel: Killed process 7376, UID 2040, (ns-slapd) total-vm:25989860kB, anon-rss:15559436kB, file-rss:12kB

messages:Sep 20 07:20:48 host2 kernel: ns-slapd invoked oom-killer: gfp_mask=0x280da, order=0, oom_adj=0, oom_score_adj=0
messages:Sep 20 07:20:49 host2 kernel: [<ffffffff81122de2>] ? oom_kill_process+0x82/0x2a0
messages:Sep 20 07:20:49 host2 kernel: [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name
messages:Sep 20 07:20:49 host2 kernel: ns-slapd invoked oom-killer: gfp_mask=0x200da, order=0, oom_adj=0, oom_score_adj=0
messages:Sep 20 07:20:49 host2 kernel: [<ffffffff81122de2>] ? oom_kill_process+0x82/0x2a0
messages:Sep 20 07:20:49 host2 kernel: [<ffffffff811227d0>] ? try_set_zonelist_oom+0xa0/0xe0
messages:Sep 20 07:20:49 host2 kernel: [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj na20 07:20:49 vovc02hr kernel: [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name