Having OpenLDAP (slapd 2.4.31) and eJabberd 2.1.5 with shared roster. Roster feel sick when it reached over 300 accounts. So I decided to migrate to fresh eJabberd.
Now I installed on test Debian 8 VM eJabberd 18.01. It work fine on test environment with couple active accounts. But when I switch it on production, eJabberd stalls! CPU load is over 100% and network traffic to both LDAP servers going up to several megaBYTES per second! It happening when users start to connect to server massively.
I got same behavior with ejabberd 17 branch.
Tried to enable caching options without success.
