To start the new year on the right foot - here some notes from experience and also based on the good will of other node runners.
Sometimes things are by design not as easy as they appear...read on

Shutting down at the wrong time can cause channels to be force closed! Downtime is bad and even shutting down gracefully can cause channels to be lost if you do maintenance at the wrong time. Be aware txs can be 'stuck' and if you are offline when they expire that can mean the channel gets force closed from the other side! Seemingly it's a design fault nobody has made public!!!

Another problem: lnd WILL fail running on a 32 bit OS when the channeldb gets too big... Be aware.
The file can no longer be compacted and there may be other problems. Nobody warns about that either. And another problem - the bolt (standard) database is pathetic. It used to take minutes to restart my node maybe 10 or more. You can upgrade to postgres and startup is just a couple of seconds!! 

If anyone is experiencing strange Disk I/O errors get in touch I may be able to provide some pointers
BTW still running KNOTS and thankful for it. Ditch core!