The attached screenshot shows that turning on the SQZ angle servo 83270, which is shown by the time when CLF_RF6 phase starts to move around, the 350Hz squeezing (yellow BLRMS3) has been better, which is what this servo is intended to do. 

You can also see that the higher frequency squeezing has been worse, and now seems correlated with the range dropping.  Early in each lock there is a time when the high freq squeezing reaches 4.7dB, which corresponds with the times when the range is highest around 156 Mpc.  This is during the thermalization, you can also see "f_s" which is the estimate of the SRC spring frequency based on the calibration lines, and kappa_c are also changing rapidly in this time, so the change in range could be explained by these changes more than the impact on squeezing.

TJ has posted a sensitivity comparison with coherence here:  83327 which shows that we have some A2L coherence to fix.