Consistent 15m Dips In Data On A Clean Install

With a fresh install and everything default I’m monitoring an interface with SNMP

  name_prefix = "OMIT_"
  name = "OMIT"
  agents = ["udp://OMIT"]
  version = 2
  community = "OMIT"
  timeout = "5s"
  retries = 1
  interval = "30s"  

    oid = "IF-MIB::ifHCInOctets.OMIT"
    name = "OMIT"

    oid = "IF-MIB::ifHCOutOctets.OMIT"
    name = "OMIT"
  urls = ["OMIT"]
  token = "OMIT"
  organization = "OMIT"
  bucket = "OMIT"

And every 15min or so I seem to be getting an odd dip that increases in severity when polling at higher rates. however this dip is not actually occurring and seems to be artificial from the monitoring.

Currently the only thing that telegraf is polling is this one snmp interface, to an InfluxDB that is only receiving data from this one telegraf.

Any thoughts? Thanks!

@Jay_Clifford any chance you can help with this please?

Hmm, that is odd, could you switch over to raw table mode and let us see the raw value of the drip? I am just looping @Hipska to this chat as well. He is our resident SNMP god. If the value is consistent we can of course omit that result using Flux or a processor plugin. However, I think it would be best to get to the bottom of the rogue value.

Polling an snmp device every 30s is very fast. Most devices don’t reserve any CPU capacity for handling snmp. So if that device is doing “something” every 15minutes, the requests at that time take a bit longer and might result in these strange results.

Could you enable the internal plugin and see the evolution of internal_gather - gather_time_ns?

I also didn’t understand what you meant with an increasing severity when using even higher poller intervals?

1 Like