Zenoss Cloud Features

  • 1.  Zenperfsnmp Discrepancies

    Posted 09-22-2021 10:41 PM
    Hi,

    We are currently running Zenoss 4 with 15 collectors located in states around the US; each containing anywhere from 500 to 1500 devices.
    Recently, we have noticed what appears on the surface to be erroneous data on some of those devices - overall, a small percentage.  In this example (see pics attached) the distribution switch (switch1 layer 2) saw approximately 40-70 % sporadic packet loss over the 24 hour period, however, the 2 switches served by, or  behind this switch saw 0% packet loss for the same period of time. As a cross reference, the device availability to all switches showed 100% availability. I understand zenping pings each device once per minute and zenperfsnmp sends 3 fpings once every 5 minutes.

    Anyone have an idea of why switch1 shows such a discrepancy?

    ------------------------------
    James Alexander
    Network Engineer
    CHICAGO IL
    4087101563
    ------------------------------


  • 2.  RE: Zenperfsnmp Discrepancies

    Posted 09-23-2021 05:18 PM
    James,

    In a previous role supporting networking hardware for $COMPANY, we saw this crop up from time to time.  For that product line, the issue was that the switch OS assigned the task of "respond to ping" with the lowest possible priority.  If the switch was ever busy with more important tasks, ping times would tank or they would go unanswered altogether, but the switch would continue to pass traffic without incident.  If you're also monitoring CPU metrics on this switch, it might make sense to compare the packet loss peaks against the CPU graph.

    I've also seen some hardware refuse pings with a 0 byte payload, but that's always been an all-or-nothing situation and isn't likely our culprit here.  If you ever find a device refuses to respond to zenping, while everyone else can ping it just fine, check the data-length option in your zenping.conf.

    ------------------------------
    Michael J. Rogers
    Senior Instructor - Zenoss
    Austin TX
    ------------------------------



  • 3.  RE: Zenperfsnmp Discrepancies

    Posted 09-24-2021 09:59 AM
    Hi Mike,
    Thanks for your reply! It appears we have resolved the issue by restarting all services on the collectors.
    Perhaps a memory leak of some desccription.

    ------------------------------
    James Alexander
    Network Engineer
    CHICAGO IL
    4087101563
    ------------------------------