Setting up OpenNMS – Part I

So, as i said almost a year a go (wow time flies) i have been dealing with OpenNMS a lot in my new job

We run a few of these servers for various things, but i have to setup a new one, and thought i would actually write as i go.

These articles will focus on deploying OpenNMS for the scenario I use it in, which is monitoring dedicated and VPS servers for web hosting clients.

How do i use OpenNMS
I work with mainly web hosting servers, but dedicated servers. So clients have root access and do as they please / need. We don’t want OpenNMS to go nuts and adde every IP in the range as floating IP’s backup IP’s etc etc come in to play. So we manually add each server to monitoring.

First things first, go and read this:
http://www.opennms.org/documentation/installguide.html

It covers how to do it so i dont duplicate how its installed

I use the RPM version through Yum on CentOS 5.8 x64 (i like CentOS)
Its also worth having a read of http://www.opennms.org/wiki/Performance_tuning before you deploy

So return here once you have it installed and can load the web GUI 🙂

I would suggest making postgres start on startup, but i would leave OpenNMS off and start it manually – why?
If your server goes down, or you suffer a bigger issue and the server loses power, you will have a torrent of alarms when things come up as everything is all over the place. So i always start it manually. Up to you.

So first thing i will be wanting from my install is SNMP functionality so i can graph things. So lets turn on SNMP logging so i can collect SNMP info with OpenNMS

First things first open up snmp-config.xml (/opt/opennms/etc/snmp-config.xml)

Here is a sample

So as you can see, I’ve done split range where you have different SNMP strings in the same range. I have also added an example of a single IP in a different community.
I have never used SNMPv3 so your on your own for that one – Sorry!

Adding nodes to OpenNMS

Now that we have SNMP done, time to add some nodes!
Click on Admin > Manage Provisioning Requisitions
Create a new “Provisioning Requisition” – these are like containers to hold the nodes. I tend to group them by the overall device (eg Network, Core network, Infra. Servers, Client Servers and then i give a separate group to each of our huge clients) these will come in later for making life easier when grouping alerts

To come in future articles:

• SMS alerts
• Email alerts
• Testing alerts
• SNMP alerts (Disk monitoring, Load monitoring, Port state changes)
• Integrating with a contact database  for a larger contact base
• Displaying customised alerts screens (eg on wall monitors)
• OpenNMS maintenance