Xymon Mailing List Archive search

SNMP Trapping Question - What is Best Tool for the Job?

list Buchan Milne
Tue, 22 Jun 2010 08:16:19 +0100
Message-Id: <user-7efb3caea0e8@xymon.invalid>

On Monday, 21 June 2010 16:54:10 user-ddebaeecde97@xymon.invalid wrote:
On Monday, 21 June 2010 13:55:53 user-ddebaeecde97@xymon.invalid wrote:
Hello;

I've gotten devmon to work on my current Xymon setup; the only drawback
in my opinion is lack of device support in devmon.
Templates are created for devices people have access to, who are willing
to do some work to create a template. I try and add all templates that
are submitted.

Even if devmon had MIB support, templates would still be required to a
degree. Feel free to assist in improving devmon.
I will try, but have found the task of creating MIBs a daunting one. Yes, I
 want to use someone elses MIB's for my devices.
You shouldn't need to create MIBs, they shouldbe provided by the device 
vendor. However, while the MIBs contain a lot of information, they don't tell 
you what data to present to the user ... so even with MIB support, there would 
still be a bit of work (e.g., the message file would have to stay). I would 
like to look at removing the need for the 'oids' file though (by MIB support).
I'd like to roll-out
something else that would allow me to receive SNMP traps from a device,
and send out alerts via Xymon.
I need to implement some trap support. The snmptrapd->snmptt->sec->xymon
method (at http://cerebro.victoriacollege.edu/hobbit-trap.html) is a bit
heavy for my current environment.

I have been wondering if a dedicated perl script using
NetSNMP::TrapReceiver (IOW, running inside snmptrapd) reporting directly
to Xymon would be better. However, the question is, exactly how should it
behave? How should traps be mapped to tests (all traps to a single 'trap'
test, or to individual tests, and how)? Should traps be stored to a
database as well (so they can be ACK'ed etc.)?
I wish I could help here, and agree that the
 "snmptrapd->snmptt->sec->xymon" is a bit conplicated, especially with SEC,
 but probably the most powerful design I've seen yet.  The problem I see
 here is the ability to create SEC rules for alerting and properly testing
 them.
Ideally, alerting should be handled the same as for other xymon events.
I'd also like to be able to use
WeatherMap, and of course want to make the three of these work together
as seamlessly as possible.
While I would like to improve Weathermap (to require less manual work in
creating map configurations), it works well enough for me

http://staff.telkomsa.net/~bgmilne/xymon/

Have you run into problems with it?
I love the weathermap feature (thank you) and is why I want to keep it. The
 only issue I have with the present weathermap feature for xymon/devmon is
 that the weathermap GUI-editor is not implemented. 
Do your network devices have CDP support? I have a cdp test that works on a 
number of devices here, and I may consider pulling that info in to the 
weathermap script, so you don't have to manually create connections between 
devices ...

Regards,
Buchan