The restart of a cluster node should work without any problems. If you encounter a problem with "authentication support service is missing", please check the error.log for CRX error messages. In case this is a production system, raise a Daycare ticket.
So, you mean first stop slave and start slave and then stop master and start master should work fine?? if so there will not be any sync problems?
Thanks in advance...
You should stop master last then start master first.
I don't have an answer about your question, but if you have to restart the CQ instance for your codebase to pick-up changes, there's something wrong in the codebase.