SNMP trap acknowledgement and more

Started by mrsackler, June 02, 2008, 06:22:56 PM

Previous topic - Next topic

mrsackler

Greetings all, I've recently been charged with choosing and implementing an open source suite for the purpose of monitoring our carrier class network, from core devices to cpe. Here is (one of) my problem(s). Please find it in your heart to excuse me if this has been discussed in another thread, as I have not been able to locate it if it does exist (or slap me with a large trout).

My first attempt has been to instruct our adtran multiplexers to send snmp traps to the netxms server. NXMS is logging all incoming traps. I've compiled all necessary mibs from our main hardware vendor into the netxms.mib successfully, matched the trap oids with their symbolic descriptions, and created events and traps to match. Here is my problem: the devices in question tend to send the same alarm traps to the server in quadruplicate.

From what I can tell, the SNMP Agent resident on the multiplexers expect some kind of acknowledgment of the first instance of the trap, else it sends it 3 more times. This is not to be confused with the acknowledgment of alarms - in fact, we want to leave alarms active until the issue has been resolved. However, we don' want 4 of each trap triggering 4 identical events, 4 identical alarms, and 4 identical emails which in turn send our whole team 4 identical SMSs for each trap at 4 in the morning. Configuring the MUXen to behave differently is not an option. If this is a capability of SNMP, how do I instruct NetXMS to acknowledge receipt of the alarms so the MUX will refrain from thrice repeating it?

Any advice is greatly appreciated.
Ryan Sackler

Victor Kirhenshtein

Hello!

What kind of acknowledgment you device need? If it can be done with some external command, you can do the following:

1. Create event from SNMP trap
2. In event processing policy, define an action for this event to call external command for trap acknowledgment.

Best regards,
Victor

mrsackler

Some of our SNMP-enabled devices using SNMPv1 provide a limited subset of options for trap handling.
For instance, you specify an snmp management server for the device to send its traps to, and retry count (up to 4) until it receives trap acknowledgment. In other words, unless we set the retry count to 0 it will resend the same exact trap multiple times because it never receives this "acknowledgment" from the NetXMS server.

It's strange, I read up on this a little and apparently only SNMPv2 or greater is supposed to expect an "acknowledgment" to its informs. Pardon my misunderstanding. But your recommendation for using the external command option in event policy is appreciated. Perhaps this particular device is strange in some way and I might be able to pass some kind of command to net-snmp via event policy action to let the device know the trap was received.