So if I'm going to set up a monitoring system for some minor hosting stuff (think ~3 servers, standard checks) – what would you recommend? I've heard about both icinga2 (which I've used, but only as a user) and prometheus. Is there a large difference in the learning curve and usability? #admin #tooltoot
@rixx Voting for #Prometheus:
- system metrics
- "is service X up" right from within node_exporter, no additional stuff necessary, same goes for "are there failed units"
- Basic monitoring running with 3 Go-binaries, superfast and supersimple setup
- scripts can just write their endstate to file and it just gets exported by node_exporter (textfile-collector FTW!)
- You have to learn PromQL, but if you do, you can really powerfully query your data, which is super-nice (and it's prtty strfwd imo)
@cherti This is what I mean by learning curve. Building a monitoring system is not my job. It's not even my side job. It's just an addendum to my unpaid sidejob.