Why quiet servers last longer
A small VPS that does one thing and logs almost nothing is the cheapest piece of infrastructure I have ever kept alive. Most of my outages came from things I added, not from things that broke by themselves.
The rule I settled on: every service gets a systemd unit, a restart limit and one line in a text file explaining why it exists. If I cannot write that line, the service goes away.
It is not glamorous, but a box that has been up for four months with nothing to report is the best monitoring dashboard there is.