4 Replies Latest reply on Mar 15, 2018 1:39 PM by mesverrum

    Slow snmpwalking (using Orion GUI tool).

    yaquaholic

      We had a WLC (one of many) which this morning decided to (wrongly) report 1/2 of its APs as down this morning.

      The network guys quickly realised that the APs were up, so blame went to monitoring.

       

      At the time, I did not know the OIDs in question, so opened the Solarwinds SNMPwalk tool and started walking the device, trying to capture the SNMP state and prove that SolarWinds is working correctly.

      That was at about 10:30am this morning.

      Some 5 hours later, the damn SNMPwalk is still 'walking' the same device.

      Here is 14:29 until 15:10, in which it did (walked) a few thousand OIDs.

       

      40 odd minutes of SNMPwalking - and still going!

       

      The APs reported up long ago, very much doubt I caught them being down, so cannot prove if it was the device or the monitoring.

      And left scratching my head as to why SNMPwalk is soooooooo slooooooow today.

        • Re: Slow snmpwalking (using Orion GUI tool).
          yaquaholic

          Dum, de dum...

          Still waiting....

           

              • Re: Slow snmpwalking (using Orion GUI tool).
                yaquaholic

                Nearly home time and some 7 hours later.

                 

                 

                Will it have finished by the morning (rhetorical)?

                  • Re: Slow snmpwalking (using Orion GUI tool).
                    mesverrum

                    I would wager that this proves at the very least that something is weird about your WLC and how it is interacting with SNMP.  380,000 OID's in a response is pretty unusual but not outside the realm of possibility for a WLC with a high number of AP's. I've done walks of devices with 80,000 OIDS that took maybe 15 minutes so your time seems slow but devices vary dramatically in how quickly they respond to polls.  To test if this is something specific to Solarwinds or not you could try to initiate a walk from another tool or Linux based machine if you have one handy. Although, to be fair I might wait until the first walk finishes before trying to throw another one at it.  Also have the network team verify that the WLC is not pegging it's cpu's or something.

                    If they both experience similar slow outputs then that points to the WLC as having an issue.

                    Do you have any other active controllers with similar amounts of AP's that you could compare results against? The only WLC's I have walked before only had like 50 AP's attached and the walk definitely did not take hours, I don't remember it seeming unusually long at all.  Not sure how many objects are on yours for a reference on "normal" times.