[ixpmanager] Odd behaviour in MRTG graphs

[Nick Hilliard]

Marinos Dimolianis wrote on 06/04/2021 14:58:
> Has anyone faced anything similar to this?

it's an mrtg problem and is known to happen in some situations, 
particularly with RRD.  The problem is that the sampling gets out of 
synchronisation with itself, so you get counter information from one 
300s period falling into the next period.

This can happen if you have a switch configured to be active IXP 
Manager, but it isn't responding to SNMP requests.

Are you running mrtg as a daemon, or running it from cron?

Nick