2 answers
- 321
regarding admin-up interfaces: NMIS calculates metrics for all interfaces that it collects, so admin-up/oper-down does negatively affect the metrics.
personally i'd avoid having known-unused interfaces in that state, not just because of the metrics but because of the potential for confusion (e.g. when being called out at 3am...). it's relatively easy to use the 'nocollect' model section to set parameters for ignoring certain interfaces, e.g. by default nmis doesn't collect ones without description text.
regarding the interaction between metrics/kpis and events: whether an event is acknowledged or not doesn't affect the metrics/kpi calculation. acknowledging just stops notifications and escalations.
in contrast, some (but not all) events' existence does factor into the kpis: for example, an interface down event will cause the node to have degraded status, thus affect its health measurement and its interface availability metric.
so, acknowleding an event won't do your metrics much good; only when the underlying problem goes away (or you tell NMIS to stop monitoring known-dead things) will the metrics change.
- Hartleigh
Thank you very much Alexander for your reply. My NMIS installation (vanilla VM) is collecting for interfaces with a blank description. Do you have a link to further details on configuring nocollect? As our routers are managed by 3rd party I am leaning towards just setting collect to no for these interfaces.
- Alexander Zangerl
sorry, i wasn't precise: *many models* don't collect interfaces w/o description text, but not all. for example, the catalyst models skip them. Our documentation for this particular aspect is incomplete, which i'm fixing right now - i'll add a link to that wiki page as soon as it's done.
Add your comment... - 10-1
i've just finished updating the documentation for controlling the non/collection of interfaces, see the following link: Controlling the Collection of Interfaces, Adjustments and Overrides
- Hartleigh
Thank you. Love your work.
Add your comment...
Hi all o/
First time poster. I have been slowly configuring NMIS8 for a few weeks and I am loving the software! Have used or tested to some degree Argent, Nagios, OpenNMS, Zenoss, Dude and PRTG over the years and NMIS seems to check all the boxes.
A quick question about acknowledging events - if I ack an event, will metrics for that event continue to be negatively affected? Or will it clear the event and metrics begin to rise again for that device? And will the status stay as degraded?
Our site routers have interfaces that are admin up but oper down and I want to keep our metrics up. Is best practice to stop collecting on those interfaces until they are actually in use? Or simply ack the event for that interface?
Thanks for your help.