From the attached picture, it is visible that there is an excess of noise in the error signals. They look similar to the ones I observed when spikes started to appear.

The spike issue is reportend in entry 428 where I also give some reference values for the rms of the mirror motion in standard condition (where no spikes are present). The rms shown in the picture attached by Yuefan is about 7 microrad which is higher that the normal one (about 1 microrad).  I'm not sure if the spikes are the only cause for the smaller range of the PR, anyway they make the local control very unstable and preventing the lock of the cavity. 

It is good that PR magnet is back to life but I'm very surprised that it seemed not respondig for a period. Indeed something strange is going on...

Pic  2 (about BS), referred to in the text, is missing.