Wanting to learn Prometheus and Grafana
Posted by Newtechintown@reddit | sysadmin | View on Reddit | 5 comments
A systems administrator job that I found for a company I’d love to work for was recently posted and they mentioned monitoring tools, Prometheus and Grafana experience specifically as a plus. Does anyone have any good home? Lab recommendations for using these tools and getting a better understanding how they work?
cabiti@reddit
Nice! I remember when I first started digging into Prometheus and Grafana. A good way to get hands-on is to set up a small Docker environment. You can run a few common services like Nginx or a database, and then use the exporters to send metrics to Prometheus. Grafana then pulls from Prometheus to visualize it all. Honestly, breaking things intentionally, like the other commenter said, really helps solidify what you're seeing in the dashboards.
H3rbert_K0rnfeld@reddit
killerkoda.com
Frothyleet@reddit
Grafana is easy - superficially, at least. You could have a Grafana server up and running in <30 min with no specific experience or hardware. Spin up an Ubuntu VM on hyper-v on your workstation, install (which is basically a one-line from their website to add their repos so you can just "apt install"), default configuration will have you functional. I think Grafana has example datasets but you can grab sample data from wherever and throw it somewhere Grafana can reference (wide variety of options but with 10 more minutes you can throw any number of DBs on that same server and upload a CSV of weather data or whatever).
Actually making dashboards and stuff, well, now you are playing with data analytics, hard to say if that will be part of your actual job.
Prometheus probably a little trickier, lot more to learn about configuring your data sources, and you'd probably need to gin up some things to monitor.
slugshead@reddit
Set up your environment. Easiest way is on docker, with a prometheus/snmp-exporter/grafana stack.
Start small and grow it, you're going to bang your head against the wall with YAML and MIBs if you're not familiar with them, will quickly make sense.
For testing and learning, start with monitoring your home routers WAN port for bandwidth use. Then bring in another metric from something else, like number of connected wireless clients.
Getting the data is the hard bit, making the grafana dashboard is a doddle, it's all GUI and rather intuitive.
I've recently made a dashboard for all our of UPS's to monitor runtime/load/output etc.
Nexthink_Quentin@reddit
spin up a couple services and then intentionally break stuff
prometheus/grafana make way more sense when you’re trying to debug something vs just staring at dashboards