10 Replies Latest reply on Jun 4, 2014 8:48 AM by shawn_b

    UDT loses device information

    Ismo

      UDT loses device information in some switches. When I do UDT discovery again, I got port information back, but within one hour it's lost again! In some other Cisco swithes with same model, same IOS and same basic configuration, device info stays as it should. Something to do about it?

        • Re: UDT loses device information
          shawn_b

          What are your polling intervals?

           

          polling jobs.JPG

            • Re: UDT loses device information
              Ismo

              At start it was default 30min but after that i've tried the other values.

               

              New information of my problem: Now devices have shown in port list for couple of days. I just don't know why! There is nothing in switch configurations that could affect to this. I made re-discovery in wednesday morning, and devices disappeared from there quite soon. When i made the same again in afternoon, they suddenly started to stay in list. Why's that? Bug?!? As told, the other switch with identical configuration is showing all the devices correctly (that other switch also lost devices once before, but now works ok) but this one needed MANY discoveries to work ok.

            • Re: UDT loses device information
              RichardLetts

              UDT is broken in so many ways that it's making me cry; I have had cases for months and they have still not fixed the code. There are fundamental flaws in the way it works (I am not talking feature requests here, I am talking that design is not working issues. ).


              check for message like this:

              2014-04-29 10:12:49,962 [ 27] WARN    -   ScheduleOrUpdateJob() - job scheduling was aborted (Capability is disabled [N:6407|Layer3|disabled|30|1].)  [SolarWinds.UDT.BusinessLayer.Polling.UDTJobManager]

               

              if it fails to poll something from a node for some reason then it decides that the node no-longer supports that functionality. For example if the retrieval of the L3 tables fails for some reason it will stop fetching L3 data and not report this anywhere (other than buried in a logfile, or a check-box on the edit-node page). Instead of rediscovery try going to the edit-node details and checking that it is polling for L2 and/or L3 data

               

              If you have Jumbo frames enabled and/or your SNMP devices return more than ~4K of UDP data then the SNMP library used by UDT will fail. There is buddy-drop for 3.0.1 (not integrated into 3.0.2 but it works for 3.0.2 ) that fixes this.

               

              This is just one of the fundamental design errors: a an assumption by a programmer in one part of the code has ramifications and side-effects throughout the whole product that make troubleshooting almost impossible.

               

              Richard

                • Re: UDT loses device information
                  Ismo

                   

                  if it fails to poll something from a node for some reason then it decides that the node no-longer supports that functionality. For example if the retrieval of the L3 tables fails for some reason it will stop fetching L3 data and not report this anywhere (other than buried in a logfile, or a check-box on the edit-node page). Instead of rediscovery try going to the edit-node details and checking that it is polling for L2 and/or L3 data

                   

                   

                   

                  Actually, after I got information to stay in table, I even disabled L2 and L3 polling from "edit node" settings, and even that didn't remove port data. So these settings doesn't seem to affect to anything in normal L2 switches.

                • Re: UDT loses device information
                  shawn_b

                  L3 settings will not make any difference to your L2 or L3 switches if they are operating at L2 only

                   

                  I had the same problem as Richard, where UDT just lost the device info, which was a UDT related problem.

                   

                  Just started to work after the job engines were reset