Using Nagios and "BEM, BMC Event Manager"'s msend.
Posted: Fri Apr 23, 2010 9:26 am
Idealy we would love to use TCP or have it exercise one of our Webservice adapters that would hand off the alerts. However, we would settle for basic SNMP traps if that was the case. Usually if no MIB file is supplied the trap defaults to a generic SNMP Enterprise trap that we can write logic around to handle those type of events. This can take some work as to some of the values that we see in these sort of traps are like 1,2,3,or 4. 1 = critical 2 = major 3= warn and 4= ok. These type of alerts will usually require us to watch them over time and cross reference that type of alert to the source system as our primer. It would be easier if there were a supplied MIB but sometimes this method is needed.As for sending SNMP traps, this would be vary easy to implement. However I'd love to hear more about how you would plan to integrate this , Nagios doesn't have a MIB file so wouldn't you have to write one for your application? Even if you were able to, obviously there are ppl who know how, why then would this solution be enterprise friendly? It sounds to me like you would be engineering your own solution from scratch using a protocol designed to work with light-wight/embedded and real time systems. This doesn't sound ideal at all, wouldn't you be better off using SSL(or plain TCP) with a custom text or XML based protocol that you would be better able to interact with?
So ideally knowing very little about Nagios we would somehow have Nagios execute a command that is provided by BEM (BMC Event Manager) called msend. This command will be passed certain variables from Nagios (ex. hostname, parameter, value, and possibly severity) which in turn will generate a TCP type message to our Event Management processing cell which we will then write logic around and integrate that into our Command Center processes and procedures.
Do you have a good reference source for our developers if they want to explore the development end of Nagios? I saw some how-to guides on the exchange but didn't know if you had anything published or tutorial on the development piece.
One thing I did notice in the Event Handling piece of Nagios is that there aren't any "Notification Types" for Up, Cleared, or Reachable events. Over the course of my experience with dealing in all sorts of types of monitoring systems, systems that don't send up notifications tend to get ignored more than systems that do send Up or Cleared events. We call them "Will Not Auto Close" events. Would the custom Event handler be able to incorporate Up type events?