My 2 cents. I had a Digital Ocean snapshot that turned out to be unbootable. (I have no explanation.) My snapshot procedure since then is I shut down all the services, power down, take the snapshot and restore from that snapshot.
Me paranoid? Yep. I also keep two snapshots, both of which have been proven to be bootable. You would think an image of a droplet that was running should be bootable, but real life can be full of surprises. Regarding rsync, at least you won't lose user data and configuration files. But restoration from rsync still sounds like work to me. Digital Ocean had one situation where a backup was bad. https://murze.be/2016/02/today-digitalocean-lost-our-entire-server/ I'm picking on Digital Ocean, but Vultr and Linode have similar incidents. If my flow was to do powered (operating) images...well I wouldn't. I'd go for rsync. I would have a script scheme to rebuild (reconfigure) the OS from a stock droplet.
