Ceph » RADOS

We are using ceph 14.2.9 with the help of rook 1.2.7 in openshift environment. We are observing that our cluster is not going into read only mode after 85% of usage though it is expected behaviour that it will do so, because of which our osd's are going down when it reaches 95 % of its total usage and hence making us difficult to recover data.Is it expected or we can conigure or tune some variables to get the expected result.

We are having 3 node osd setup with each node having 1TB as data and 50Gb as metadata volumes attached to it. We are currently working on openshift 3.11 on top of which rook is deployed.