Today I found that nine days ago, the H1:PSL-ISS_SECONDLOOP_CLOSED channel was OFF, so the IMC_LOCK guardian could never close the ISS second loop.
TL;DR I left H1:PSL-ISS_SECONDLOOP_CLOSED ON (one), so if we can't lock please turn H1:PSL-ISS_SECONDLOOP_CLOSED back to OFF (zero) 

I engaged the loop by hand this last lock, turning on H1:PSL-ISS_SECONDLOOP_CLOSED then using the IMC_LOCK guardian states. 
There was a huge transient in the diffracted power, from 2% to 9%, which caused a large drop in the input power.
The diffracted power usually slowly returns to it's appropriate value, but in this case it did not, returning only to around 6.5% diffracted power.

I'm very confused by why this should be the case.  
It seems as those the H1:PSL-ISS_DIFFRACTION_AVG is in loop, being input to the Slow Offset Servo.
Does some math on H1:PSL-ISS_DIFFRACTION_AVG, including an integrator in H1:PSL-ISS_SECONDLOOP_REFERENCE_SERVO, but the output (H1:PSL-ISS_SECONDLOOP_EXC_MON, orange in the plot) just grows and grows continually.
If this output was actually making it to the loop, it should have caused the H1:PSL-ISS_DIFFRACTION_AVG to lower until it was near 2.2 percent, which is the offset set in H1:PSL-ISS_SECONDLOOP_REFERENCE_DFR_CAL_OFFSET.

I looked through the actual simulink model at /opt/rtcds/userapps/trunk/psl/h1/models/h1psliss.mdl and it appears the secondloop MEDM screen is not misleading in any way that I could see.

Other weirdness:
1) ISS QPD seems to be poorly aligned
2) Power on the ISS array PDs is extremely unequal (likely related to (1))

Reference alogs:
I engaged ISS second loop this year: alog 61697
Jenne engaged ISS second loop this year: alog 61495
Daniel instructions for second loop engagement: alog 58845