[TriLUG] OpenNMS questions

[Tarus Balog]

On Sep 1, 2005, at 3:28 PM, Owen Berry wrote:

> With the recent discussion about Nagios and OpenNMS

I can answer the OpenNMS questions (grin).

> * Can it be used to monitor non-network/service related items? For
> example, a batch job that is supposed to be completed by a certain  
> time.

A qualified "yes". Any NMS is only as smart as the information it is  
able to access, thus there would have to be some method for  
determining if your batch job was "up" or "down".

One method would be to configure an agent, say, net-snmp, to "test"  
your service, and then use OpenNMS's SNMP poller to verify that it is  
"up" or "down".

Or, you could have the same or a similar agent send events into  
OpenNMS. There is a nifty little perl script called "send-event.pl"  
that comes with OpenNMS that you can use for this, or to avoid agents  
totally you could instrument the application running the batch job to  
send events to OpenNMS.

> * Related to the above, Nagios uses a plugin type system that  
> allows you
> to monitor almost anything. Does OpenNMS have something similar?

OpenNMS is written in Java, and there is a link on the main  
opennms.org page to an article on IBM DeveloperWorks website that  
goes over, step by step, how you would write a plugin.

Or, you can write any script that will run on the OpenNMS server and  
use the GPMonitor (General Purpose) to have OpenNMS run the script  
and parse the output to determine success.

Finally, the new client has an NRPE client that will plug into the  
Nagios agent and return that information as well.

> * Can OpenNMS receive and monitor events from other servers? For
> example, a server behind a firewall may need to contact the OpenNMS
> server, rather than the other way around. In Nagios terminology, this
> would be a passive monitor.

Sure. One of our clients likes the notification system so much, they  
use it to handle support e-mail. Someone sends an e-mail to  
"support911" and .procmail parses it, sends an event to OpenNMS via  
the send-event.pl script, and it gets sent to the proper parties,  
complete with escalation.

> * Are you able to configure OpenNMS for distributed monitoring and
> automatic failover?

Yes and no. Our distributed system is called OpenNMS 2.0, and it is  
probably a year out. It is very easy to have a failover system, but  
it is "near line" and not automatic. It is also currently possible to  
have a number of OpenNMS systems doing monitoring and talking to a  
central database, which is somewhat distributed but much less than we  
plan to have.

Feel free to join the opennms-discuss list on Sourceforge if you want  
a more unbiased description. (grin)

-T

-----

Tarus Balog
The OpenNMS Group, Inc.
Main  : +1 919 545 2553   Fax:   +1 503-961-7746
Direct: +1 919 647 4749   Skype: tarusb
Key Fingerprint: 8945 8521 9771 FEC9 5481  512B FECA 11D2 FD82 B45C