That underlines my thoughts on my recent suspension issues down to ~70%, related to issues mentioned in this post: everything which could block db reads/writes can cause damage to the node somehow. In my case Time Machine backups on MacOS; before that, SMB issues as well.
Nevertheless, I use all these learnings and failures to extend and tune the health check monitoring script - I strongly advise to use that, in case you don’t have a monitoring tool to recognize errors yet.
I made the experience it makes you nervous, when alerts are fired, but you can react very quickly, before seeing your node being suspended or disqualified: