We are on mp 4.0.2 with postgres DB as repo and about 65k users. Each having aprox 20-40 assignments. Running attached tasks depletes all available OS resources (cpu eventually on 800%, mem at 95% out of Xmx=16g on jdk11). This happens around after 35-40k users were processed in the task. When resources are depleted task continues extremely slow.
Pls advice, we havent introduced caching and I also may try switch to http://midpoint.evolveum.com/xml/ns/public/model/iterative-scripting/handler-3 handler. But Im curios why this is happening since I thought these kind of perf issues were fixed in 3.9 (e.g. with aggregateOutput=false).