Monitoring Network Traffic on Windows Server seems inaccurat

[chrisbooth]

I have NSClient and SNMP turned on a windows server

using NSClient for CPU usage / uptime / disk space etc but i'm not aware if you can monitor the bandwidth of the ethernet ports using it.

so i've turned on snmp and using the NagiosXI Configuration Wizard: Network Switch / Router i've added it this way (i assume using SNMP)

I'm getting data but when i'm compaing it to the task manager on the server itself it's not the same as the graph in nagiosXI

is this because the SNMP check returns Octets and it's being displayed at Mbps? Am i missing a step?

check_xi_service_mrtgtraf!10.253.0.57_12.rrd!500.00,500.00!800.00,800.00!M

$USER1$/check_rrdtraf -f /var/lib/mrtg/$ARG1$ -w $ARG2$ -c $ARG3$ -l $ARG4$

i know i probably shouldn't be pretending the windows server is a switch/router and using that configuration wizard but i wasn't sure how best to ad this service to my host other than that one

[chrisbooth]

also, i've only ever used active checks but i'd quite like my network bandwidth from servers to be updated quite often, say every 10 seconds

would it be wise to then have that set up as a passive check?

[tgriep]

The bandwidth info that is displayed in XI is not live so there will be differances between that and live data.
The XI server uses an application called MRTG to sample the information from the devices every 5 minutes.
Then the check_mrtgtraf plugin is run, looks at the data that was captured by MRTG and displays it and that is done every 5 minutes as well.
Depending on the timing, those 2 processes will not happen at the same time and that is why the data looks different from live.

Does that help out?

I haven't heard anyone doing passive bandwidth checks.

[dwhitfield]

To add on to what was already said, if you need detailed network traffic information, you may want to take a look at NNA: https://www.nagios.com/products/nagios- ... -analyzer/

Also, it looks like you can do bandwidth stuff with performance counters, but it would take me some time to figure out how exactly to do that. From what I've read, performance counters are useful as a last ditch effort, but are buggy, so not a good first choice. Now, whether it is the performance counters themselves of the way NSClient deals with them, I couldn't tell you. Maybe that will help point you in the right direction though.