Dave, Jonathan,

We restarted all the tonight starting at 16:30-17:40.  This was due to unexpected model behavior where epics outputs did not match daqd data.  We are working to understand the mechanism.  Our supposition is that it was due to a slow /opt/rtcds on model startup.  We did not need to restart any daqd processes, which points to this issue being internal to the models.  Dave will fill in a few more details in a comment.