Re: [Devmon] T or even P bits speed output in 'home made'-cisco-n7000 switch template

[W.J.M. N. <Wim...@nl...>]

Hello Peter,

> I currently want to monitor our cisco n7000 switch using devmon.
>
> I took the current cisco-n6000 template, since it is using the 10GB
> Highspeed ports and our 7000 switch is using many of these 10Gbps ports. It
> is also based on this article with suggestions from Joshua Krause:
> https://sourceforge.net/p/devmon/mailman/message/19631280/
>
> The good part: It does gather all kind of data. ;-)
>
> The bad part: It shows over time many spikes whereas the scale is boosted
> to T of even P-bit, making the output less useful.
Are the occurrences of these spikes related to certain events?
A spike can be caused by a reset of the SNMP-counters. If the DS type is 
COUNTER, a huge spike is generated, as the reset to zero is interpreted as 
a counter wrap. In case of a 64-bit counter, the result will be in the 
range you describe.

(Note: Devmon does not know if a counter is 32-bit or 64-bit. If the value 
decreases from one sample to the next, and if the value of the oldest of 
the two samples is less than 2^32, it assumes that it is a 32-bit counter. 
Thus the height of the spike will be much less for a rarely used interface.)


> Perhaps this has to do with the template that was copied, but perhaps it is
> just just the flapping from the maximum number of positions.
> (32bits/64-bits). But then again, I read 64-bits counters.
>
> I try to add ifInErrors, to remove redundant interfaces that were
> previously shown. Lending this from the if_err-test the number of interface
> shown reduced to the same amount that is shown with if_err and if_dsc-test.
> But underneath the output for all devices is shown to devmon and perhaps
> that makes it want to update for example Ethernet4_6 twice?!
Do you see the interfaces multiple times in an snmpwalk of ifName (OID 
.1.3.6.1.2.1.31.1.1.1.1)?

Regards,
   Wim Nelis.