Monitoring Total Errors on a Router Interface

Question

Hi,

I'm having a little trouble setting up some alerts for my company.  The way that we want to setup the monitoring and alerts for interface errors, CRCs, etc. is not in/out errors in the past hour or 24, but total errors since counters have been cleared (these are all cisco routers/switches).  So far with NPM I've been able to setup the monitoring that I needed with little trouble, however I'm not sure if there is a way to do custom polling like this, since my needs have been fairly basic so far (up and down alerts, running reports, etc).  Any help would be appreciated.  Thanks.

-Garrett

DLI-Alford · Answer

Hi All,

thank you for your input and help on this.  Originally I was just setup to trigger for errors in the last hour (because its easy to setup and still tells me what I wanted to hear) but my managers had to be difficult =P  I will refer them to your answers and just set this up as I had originally planned =)

FormerMember · Answer

ok this told me what I wanted to hear.

You want to know as soon as they start building up.

If you trigger based on the last hour this will give you the alert you want.

so set your trigger for >100 errors  in last hour and i am pretty sure you will get the results you want.  Becasue this will trigger the moment that the count goes over 100 and not wait till the top of the hour if that was what you were worrying about.

Did this for all Ethernet interfaces for the Net Engineers and they are able to find mismatch duplexes and bad fiber SLPs really fast

aLTeReGo · Answer

I think the issue you'll face creating Universal Device Pollers for this purpose is that the Cisco "clear counters" command does not reset the SNMP counters, just the counters you see via the CLI. The only way to clear the SNMP counters on a Cisco switch that I'm aware of, is to reboot the switch.

Even if that wasn't the case this could be done using the Universal Device Poller, but it would be a total pain in the ass to setup. For instance, if you wanted to monitor receive errors for interface 5 on a Cisco 3750 switch, you'd need to create a UDP to monitor OID 1.3.6.1.2.1.2.2.1.14.10005. To monitor transmit errors on the same interface, you'd need to create a UDP for OID 1.3.6.1.2.1.2.2.1.20.10005. As you can see this doesn't scale real well. For a 48 port switch you would need 96 UDPs to capture the information you're trying to collect.