Hi James, I think the best way to go about this is to stick to a default (per second) but have a parameter that allows that to be changed. I say that because whenever I add a new SNMP based check, I *always* verify all available parameters and tweak them to my specific needs. The only thing you might want to consider is append a "_Xsec" to the perf_data name to reflect the frequency being used to keep graphs with different frequencies separate.
Rafael On Wed, Jun 30, 2010 at 2:22 AM, <[email protected]> wrote: > > Hi, > > What I did in the "old" days was sampling per 2 minutes. The amount of > snmp traffic generated when the sample time was lower, became too much for > all devices. For those links which required it, I added the max value on > the line, to see where averaging caused misinterpretation of the > utilization of the links. > > hth > paul > > > > > > A question to SNMP users... > > > > If you're monitoring a SNMP counter, what timeframe would be > > appropriate for calculating rates? Eg: (where N is timeframe) > > > > 1 error per N > > 1 discard per N > > 1 fragment per N > > 1 degree (C / F) temperature change in N > > > > > > Setting N = seconds seems like a natural choice but I think this will > > be problematic if you're working in terms of minutes or hours. > > Would setting N = minutes be counter intuitive? > > > > > > -- > > James > > > > > > > > > > _______________________________________________ > > Opsview-users mailing list > > [email protected] > > http://lists.opsview.org/lists/listinfo/opsview-users > > > > > _______________________________________________ > Opsview-users mailing list > [email protected] > http://lists.opsview.org/lists/listinfo/opsview-users > -- Rafael Carneiro, BEng http://ca.linkedin.com/in/rcarneiro
_______________________________________________ Opsview-users mailing list [email protected] http://lists.opsview.org/lists/listinfo/opsview-users
