Yep, we needed to restart the endpoints to get some new code in there.  Should fix problems for people who were having disconnect/reconnect issues, but took a few minutes of downtime, so during that time the pool hashrate plummeted (its over the last 5 minutes), and should now be sliding back up to normal