monitoring interface errors/collisions

Question

I want to monitor interface errors and collisions but can't find the correct OID for this, but this is what I think will provide me with the correct output:

5 minute input rate 10000 bits/sec, 7 packets/sec
5 minute output rate 142000 bits/sec, 24 packets/sec
   143297898 packets input, 34571883835 bytes, 0 no buffer
   Received 2732090 broadcasts (295488 multicast)
   0 runts, 0 giants, 0 throttles
   0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
   0 input packets with dribble condition detected
   253560814 packets output, 117274967390 bytes, 0 underruns
   1341799 output errors, 542123 collisions, 0 interface resets
   0 babbles, 259128 late collision, 0 deferred
   0 lost carrier, 0 no carrier
   0 output buffer failures, 0 output buffers swapped out

-> input errors: 1.3.6.1.2.1.2.2.1.14 ( ifinerors )
-> crc: ?
-> output errors: 1.3.6.1.2.1.2.2.1.20 ( ifouterrors )
-> collisions: 1.3.6.1.2.1.10.7.2.1.4 ( dot3statssinglecollisionframes )
-> late collisions: 1.3.6.1.2.1.10.7.2.1.8 ( dot3statslatecollisions )

BUT, these are incremented figures, I'd like to have an average per second or a way to see how many of these errors occur on a timed measurement ( like x/hour or x/day )

edwardle · Answer

These MIB values are counters.  Leave the unit and timeframe fields blank.

Once the poller has been created, you can then add a custom poller chart to the orion web site to graph your data.

daxvancamp1 · Answer

thnx for the information, but i have an additional question.

How am I able to setup this monitoring to be an incremental view on a daily basis ?

I have no clue what to use in the Mib poller properties

Rate ?
Counter ?
unit ?
timeframe ?

What's the exact explanation for these properties, it's nowwhere explained at all.

edwardle · Answer

To monitor CRCs, you could try monitoring 1.3.6.1.2.1.10.7.2.1.3 (dot3StatsFCSErrors) from the ETHERLIKE MIB.  More details on this MIB:

tools.cisco.com/.../BrowseOID.do

I generally find the out of the box interface monitoring is OK which just monitors transmit errors, but then again, all my interfaces are full duplex, so I'm not interested in collisions and generally any errors I see are CRC.

Remember that collisions are normal for a half duplex interface, especially if there are many other devices on the same segment.  If this interace is not meant to be half duplex - check that you haven't got a duplex mismatch.  It is not unusual for ports set to auto speed and duplex to negotiate incorrectly.  I always force the speed and duplex on the port. The only reason I've babbled on about duplex is that it's not that common to see half duplex interfaces and collisions these days in network infrastructures, unless you've got some really old kit.

Hope this helps.