SNMP Data Collection Failure Alerts?

I have noticed that when SNMP Data Collections stop functioning on the nodes for which I am collecteding data via SNMP, Orion doesn't send me an alert.

This type of event is important as I need to know when I am no longer getting important data from my nodes. Is there any way for me to get Orion to notify me when this happens?

Thanks in advance!

Find more posts tagged with

Accepted answers

bshopp

If you have NPM 10.1 you can use the new advanced SQL Advanced Alert feature, see here and here

All comments

FormerMember

I second that. It would be nice to be alerted if SNMP collection stops functioning for whatever reason.

mrxinu

How do you mean stops functioning? The poller completely falls over or a batch of nodes aren't reachable via SNMP any longer?

byrona

Often we have nodes (more often Windows than anything else) that will stop responding to SNMP polls. This seems to happen to a small handful of nodes after each patching window for reasons that we can't explain.

Our current NMS will let us know when these data collections have failed. I would like to have the same sort of functionality with Orion and was trying to figure out the best way to accomplish this.

mrxinu

Here's the thing - every node has a last polled DATETIME column in the Nodes table. Even if that were exposed in the alert interface, there's no way to do any kind of date comparison to determine how "stale" the time stamp is to trigger an alert.

That being said, you could have a scheduled task in the database that fires once a minute (or longer, depending on the number of rows in your Nodes table) that looks at the LastPolled value (I'm not sure what the field name actually is), compares it to the current time, and then writes that delta in a custom property (just another field in the Nodes table). You could then have an alert check that value for a number higher than X seconds/minutes/whatever.

byrona

This is an option, in fact we are doing something similar to this on a more global scale to continuously make sure our polling engine is running properly as we had a lot of problems with it stopping for no reason a few months back.

That being said, I guess I would expect this type of functionality to be part of an NMS. Most other NMS's I have worked with (OpenNMS and OpenView most recently) were able to let me know when my nodes were no longer being monitored as I had set them up to be, specifically with regard to SNMP data collections. Having a system that is designed to monitor alert you when for some reason it is no longer able to do it's job seems fundamental.

If this functionality is not native to the system in some way I would like to submit it as a feature request.