I am attempting to graph simple queues with MRTG, but the outgoing interface graph shows a spike at the start of each 24 hour period. This makes the graph hard to read, because of changes the scale. I'm not sure where this is coming from or how to mitigate it. I've tried using different SNMP version, but the spike remains. I suspect there is a counter rolling/resetting that is causing this. Does anyone know what this might be?
Here is an example MRTG config I'm using:
Target[10.0.1.10]: .1.3.6.1.4.1.14988.1.1.2.1.1.8.1715&.1.3.6.1.4.1.14988.1.1.2.1.1.9.1715:public@10.100.0.2:::::1
Options[10.0.1.10]: noinfo, growright, bits, nolegend, nobanner
SetEnv[10.0.1.10]: MRTG_INT_IP="10.0.1.10" MRTG_INT_DESCR="10.0.1.10"
MaxBytes[10.0.1.10]: 1250000000
Attached is a picture of the graph.
You may wonder why I'm not using the router's built-in graphing. In short, I am. However, I have just shy of 1000 queues. Mikrotik support informed me that the flash was not designed to graph this many interfaces. Indeed, I've seen corrupted flash on three routers so far and others will sometimes stop graphing until a reboot. Mikrotik said that graphing this many interfaces requires a separate appliance. This is what I'm attempting to do. Move graphing off the router onto a separate appliance to avoid flash issues.