Customer runs two similar workloads during different periods on the same cluster,  one with mirrored reads at 1.0 and the other at 0.5 sampling respectively.  1.0 does not yield the expected mirrored reads operations on the secondary.  0.5 works as expected.