Borja has continued working on his analysis of signals sent to the suspensions after locklosses, (see 23831), he is working on another alog with the current results.  

After his original log, I made some changes to the ISC_LOCK down state to make it more efficiently turn off loops, and removed some things that were shutting of DRMI loops in the ISC_LOCK guardian, becuase those loops are also handled in the ISC_DRMI guardian.  However, what was done in ISC_DRMI wasn't turning everything off.  Today we tried to make similar edits to ISC_DRMI to make sure it was efficiently turning off ASC and LSC feedback to the suspensions.  We have tested this once, and fixed one error.  The new code is very slow to complete, ten times slower than extremely simlar code in ISC_LOCK, and takes a total of 19 seconds to finsh turning things off.  Since we don't understand why this code should be slower, we filled a bug 1068  The currrent svn revision of ISC_DRMI is 14928 

For now we are leaving the new, slow code in place because at least we think it is better than the old code which was also slow.