While watching operators relock today there were two things that we changed to help make future re-locking more reliable.  (This wasn't commissioning, we just addressed some existing problems with the locking while we were relocking).

New state for green arm guardians:

• We added offloading of the green WFS to ETM +TMS a few weeks ago.  This morning Travis encountered a situation that guardian wasn't set up well to handle, where the arm unlocked while it was offlaoding.  The offloading state doesn't check for locklosses, since we want it to complete the offloading even if the arm unlocks.  However, since ISC_LOCK was requesting the offloaded state, the arm guardian never went through down and relocked.  I added a new state, called ETM_TMS_WFS_OFFLOADED directly after the offload state in the graph, which is now requested in LOCKING_ARMS_GREEN.  This state checks if the green arm is still locked, and will respond appropriately if not. 
• One note: we still need to look at the offloading sequence, which doesn't seem as smooth as it should be, although we didn't have time for that today. 

Offsets in AS36 for DRMI ASC:

• We have a long standing problem that AS36 has offsets that seem to change.  We first engage the DRMI ASC with the beamsplitter controlled by AS45, then after the other loops have all converged we change over to 36 before doing the CARM offset reduction.  But DRMI usually becomes misaligned when we change over, and sometimes this is bad enough to cause locklosses.  After today's first lockloss we tried putting offsets in the AS36 Q pit and yaw filters after the ASC converged with the beamsplitter controlled by AS45,  switched over to 36 for the CARM offset reduction, and turned off the offsets before engaging the full lock ASC. Since it worked, we added it to the guardian to set the offsets based on an average of the AS36 value once the ASC has converged.