Nagios Monitoring & Branches of?

[BongoHunter]

You need plugins, and then you need to define checks to use the plugins

Good start is a host check that tells you if agent can talk to icingan, then get some basic perf monitoring in place, with CPU, memory and disk utilisation/IO counters / graphing.

Then add service checks for the core windows services. Think about things that should run on every host you have an agent on at this stage (like backups, failed scheduled task, eventlog etc)

That should cover you for the basics of host and OS, stuff beyond that will be specific to server role. So you probably want a base template in icinga for all the standard stuff, then templates that build on that for servers with specific roles.

 

[DHR]

Cheers BongoHunter, made some progress it's just slow. I suppose once it's nailed down rollout will be so much easier though.

Managed to get the poweshell module talking to icinga, so it pulls down the agent and nsclient++ installs and does some config that's working pretty well. It's just getting the plugins and checks in place now.

Because I've jumped straight into using icinga Director I think my understanding of the nature of the plugins has suffered a little.

 

[DHR]

Made some awesome progress at last with icinga2 and director, feeling a lot more confident and can start rolling agents out soon.

Trying to get my head around a decent way to capture cpu usage over time, but aside from that i'm more comfortable than before.

Just contemplating looking at getting some graphing stuff installed ideally with a bit of data retention too.

 

[DHR]

So looping back round to total annoyance again!

I've looped around a handful of solutions for data retention and graphing, was strugling to get graphite going for some reason, but InfluxDB and Grafana seemed perfect. I've now hit a wall with that whereby I'm not convinced I can grab the statistics from things coming from nscp losing the will!

Everything I've read this afternoon is pointing toward using telegraf instead for windows hosts Soon as I box something off and feel like I've made progress, another 10 open!

 

[Conanius]

Icinga2, Grafana and Influx DB is what we used where I used to work. I can share this thread with the guy who designed it if you want pointing in the right direction?

 

[LizardKing]

Zabbix, NXlog agent and grafana. Grafana isn't needed as Zabbix is good at graphing on its own but Grafana makes them pretty if that's what your after

Though if your already mentally invested down the Icingna rabbit hole then Zabbix is not likely to appeal I thought I liked Nagios till I setup Zabbix. (credit to Nagios, I've still got servers that were setup 10+ years ago that still tick over much like the robot in the film silent running!)

 

[DHR]

Icinga2, Grafana and Influx DB is what we used where I used to work. I can share this thread with the guy who designed it if you want pointing in the right direction?

Yeah go for it that'd be cool. I'm pretty much there with it now. Biggest disappointment really is the CPU monitoring with nsclient for Windows not being able to take samples without you running the webserver in each agent. Need to look at different ways to capture that.

It's been (and will continue to be) a struggle to get to grips with.

I did look at Zabbix but I went toward icinga based on install base size and that stupidly thinking the community support etc would be better.