So, Sensu is a bit of a flop. It’s got a great architecture, but it’s so fidly, badly documented and makes you feel like a complete idiot half the time when you try and configure Slack alerts to go to a different channel.
Meanwhile, their website seems to be completely dedicated to selling you their “enterprise” version and it’s written in Ruby.
So… the main reason we didn’t go with Prometheus was because it seemed that Sensu would be easier to get up and going, being closer to the “Nagios” way of doing things (i.e. making it super simple to add simple checks). However, it seems that with the node_exporter we can get filesystems metrics, cpu, RAM and process metrics etc… which might be a go.
So, the plan is:
- Spin up a Vagrant Ubuntu box and get Prometheus running on it
- Get Prometheus monitoring its own metrics
- Spin up Vagrant Ubuntu box and get Grafana running on it
- Spin up Vagrant Ubuntu box and get Prometheus Alert Manager running on it
- Document the whole thing
- Do the same thing in MiniKube with containers
Let’s see how this goes