It seems that our trouble locking last night was due to a change in the initial alignment references.  We are now locked at DC readout with a recycling gain of just over 30.  

In order to lock I changed the TR_CARM offset (sqrt(TRX+TRY), each normalized to single arm power) at which we transition to the REFL PD from -41 to -35.  This allowed us to lock with a recycling gain of just over 20.  After that I moved the ITMs by hand with all the ASC except the soft loops on, and we could recover a good recycling gain with the old TMS offsets.  That means that the problem was just our initial alignment references. 

At DC readout, I used the scripts described in 37033 to reset the green alignment references, although I had trouble getting the Yarm to lock in green. Screenshot shows the changes in settings.