We could either expand our current cleanup job to take extra safety measures (such as making sure that the principal has been gone for X amount of weeks in IT before removing it from RBAC). 

This could be achieved by adding a field in the principal table for being inactive and writing logic around it to check for the date/ check that it was already marked inactive once before to remove it. 

Or we could investigate moving to track messages on IT's UMB about user activities (such as deletion) and moving to an async task. 

NOTE:

Clean up job calls BOP proxy for each principal here.

https://github.com/RedHatInsights/insights-rbac/blob/5610ebadc9c92f8d6fbf5051eeb34a00969f6457/rbac/management/principal/cleaner.py

we need to investigate where there possibility to reduced number of calls to BOP and it endpoint - this could be also solved with IT's UMB.