If I am reading this correctly, the change being implemented is increasing from 90% to 95%? I'm not sure that's going to address the issue. I believe the values for the actual system reserved need to be increased, not the percentages for the alert. Possibly I am misunderstanding the full change but most of the clusters I've seen were well over 100% of the default values being utilized.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (OpenShift Container Platform 4.7.33 bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:3686