Follow

what do you use (not at work) for monitoring servers?

there's a zero percent chance I want to set up a full ops monitoring stack, but that's all I'm really used to dealing with.

I'd just like to ship logs and monitor main metrics like system load, network usage, disk usage, heavy procs.

@adasauce Just as a data point, I'll share that I'm using #Opsdash but would like something different. The configuration UI is a clunky and it's got this freemium-style pricing plan where it's free for a limited number of servers and services. Documentation is basically nonexistent.

@annika ah yea I've seen opsdash a few times, and for some reason get a lot of ads for pulseway.

on further reflection, I think I want to do something more compose-able where I can build on and extend to monitor a bunch of different stuff.

that obviously takes it from simple stress-free non-work monitoring though.

thanks for the recommendation, really made me think more about what I was looking to do.

@adasauce At home, I'm still mulling what I want to use for logs. For metrics, I use collectd on hosts with the write_graphite plugin to ship and store metrics in graphite, and Grafana to display. I love Grafana! Then Grafana alerting uses webhooks to send notifications to Mattermost.

In addition to collectd, I have some SNMP and some custom scripts. Also Saltstack (for consistency) and trying out Sensu.

@adasauce I'm personally planning to use Prometheus in combination with Graphana but that sounds like more than what you're looking for. There's also netdata but I don't remember if it can aggregate metrics from multiple servers.

@amolith following up to let you know that I've deployed:

prometheus
loki
grafana
alertmanager
node-exporter
cadvisor
promtail

I have a few dashboards set up, and it's going pretty well so far. A+ recommendation.

In the end I couldn't get over how commercialized the elastic stack was, open core is just vomit inducing.

@adasauce Historically: Munin. It is lo-fi (but not in a good way), but is super simple to set up.

Have been planning to look at Cacti, but only since I already have Ansible roles to set up all the extra crud it needs (web front end w. PHP, MariaDB).

@adasauce zabbix, but im not 100% happy with it.

Call it a luke-warm recommendation.

@smallsees yeah, I use zabbix at work a lot.

I'd really like to use something that's not zabbix for at least a change of pace.

I appreciate the recommendation though :)

@adasauce We've just started testing ELK at my work for aggregating logs. It also reports resource utilisation, and looks really cool these days. We used ELK for a couple client projects before and it has come a long way since then.

You didn't mention monitoring service availability and functionality, and alerting on those things, though that's what our main monitoring systems are used for at my workplace. For that, we have been using Nagios for years and it works really well. :)

@michelamarie interesting! I haven't used ELK in probably 5 or 6 years.

This might just be the perfect solution to my problem. Ship logs, ship metrics, looks like it (or via integrations) can do alerting based on those metrics too!

@adasauce Ya! It's totally worth a try. It's been around five years since I last used ELK as well, and it really seems to have improved in that time -- better components, more straightforward installation, and a bunch of new features.

Sign in to participate in the conversation
Radical Town

A cool and chill place for cool and chill people.