14 Replies Latest reply on Jan 20, 2016 6:04 AM by kleong

    Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2

    Bradley_M Apprentice

      We are having major issues in software delivery where it is literally taking 20+ minutes to decide if a machine is off for each machine.  Does anyone know where this setting is?  We are running 9.6 SP2 and am not able to find the setting.  We didn't change any of the defaults.

       

      Bradley

        • 1. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
          nick.evans SupportEmployee

          Hi Bradley,

          In 9.6 agent discovery is attempted repeatedly to see if the device is available. If you were to watch the traffic for it, you would see multiple discovery attempts/UDP Packets go out looking for the agent.

          In your scenario of 20+ minutes, are you indicating that they will eventually discover after 20 minutes, or that it takes 20 minutes to decide they are off?

          The duration a task takes can be relative to the # of devices on a task as well. For example, a task with 1,000 devices will try all of them before retrying them all.

          Does this issue occur on every task no matter what?

          Does the # of devices on the task make an impact?

           

          This article outlines the discovery process and offers some good starting points for troubleshooting potential issues.

           

          How to troubleshoot Agent Discovery

          • 2. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
            Bradley_M Apprentice

            Hello Nick,

             

            It will take 20+ minutes to decide if the machine is off.  The most extreme example is I built a managed script and attempted to deploy to a machine that was purposely off.  The task ran for multiple days and you couldn't cancel it.  I had to restart the global scheduler to kill it.

             

            The 1000+ targets happens to me often and I see this problem the most with large target groups.  This is very bad for me if it is continually trying off machines for the duration of the task.  If it is only targeting about 60 machines at a time to get through that 1000+ it will get saturated with machines that are off and will slow the task down to nothing.

             

            Is there a way I can adjust it to operate like the previous versions?  Our 9.5 cores can chug through 1000+ machines easy but on the 9.6 SP2 core it takes 3 or 4 times longer because it is constantly trying to hit machines that are off.

             

            Bradley

            • 3. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
              Kenyon Expert

              We are having the same issue here. I have had a case open with support for well over a month with no answer. This happens almost solidly when I send jobs to a group of laptops where some machines might be off or unreachable. The whole job will eventually fail all of the targets with "cannot find agent" even though many machines are pingable. So far support seems to be leaning towards agent discovery but as in your environment, the 9.5 server works just fine.

              • 4. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                nick.evans SupportEmployee

                Hi Kenyon,

                In 9.6 we added secure detection where we ask clients for certain information about themselves before we allow them to be considered the correct party. In 9.5, whoever responded to the discover at an IP got the job. In 9.6, if a new machine responds to the IP discover (perhaps IP's were recycled) LDMS won't give the responding device the job which is a good thing. I would recommend going through the agent discovery article to look at things that could be impacting your particular issue.

                 

                How to Troubleshoot Agent Discovery

                • 5. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                  nick.evans SupportEmployee

                  Hi Bradley,

                  You could enable accelerated push which customers have had success with because it effectively will try more machines at once. The maximum run time can be defined within the Scheduled task settings (scheduled tasks | Settings (gear icon)), however, it has a minimum of 30 minutes.

                  https://community.landesk.com/support/docs/DOC-35474/#jive_content_id_Enable_verbose_policy_task_handler_logging

                  • 6. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                    Kenyon Expert

                    I have already tried accelerated push as well as changing the maximum run time to 60 minutes with no change in results.

                    • 7. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                      Kenyon Expert

                      I was given the How to troubleshoot the agent discovery doc and have thoroughly reviewed it. Support is reviewing wireshark trace files now. The 9.6 secure detection would make sense if the job would fail on some machines but the task ultimately fails all target machines. My jobs that target mostly desktops work better and most devices succeed.

                      • 8. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                        nick.evans SupportEmployee

                        Hi Kenyon,

                        The accelerated push recommendation is for the OP. Where you can ping devices but not discover them, you should go through that article and try the steps about real time discovery.

                        • 9. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                          Bradley_M Apprentice

                          Hello Nick,

                           

                          I have tried the tasks with Accelerated Push and it operates a little better.  I don't want to set the maximum scheduled task time because it might kill the task before it got through all of the targets one time.

                           

                          This will not help me on my managed script deployment either because there is no such thing as an "accelerated managed script deployment".  You really can't set any settings if you are deploying one.

                           

                          Also something extremely irritating is that these machines are coming back as "Failed" instead of "Off".  Now I can't set a task to repeatedly target machines that have "Not tried to run the task" because it considers a machine that returns a "Failed" "Cannot Find Agent" as already run the task.  It doesn't operate that way in 9.5 either.

                           

                          Bradley

                          • 10. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                            Kenyon Expert

                            Bradley,

                             

                             

                            Have you applied any post SP2 patches? I personally do not recall this issue prior to installing one of the post SP2 patches. I cant remember if it was after 722 or 812f that we noticed this after.

                            • 11. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                              Bradley_M Apprentice

                              Hello Kenyon,

                               

                              We installed comp0715 and comp0815 but that is the patch level we were at when we started using the cores.  Our move to 9.6 SP2 is recent and that is what we built our cores out to.

                               

                              Bradley

                              • 12. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                                Specialist

                                Same issue in our environment when pushing out task. Put 100 pc's in they will all try to discover the maximum task run time than all fail. Nothing is reaching the clients. When running a small number of computers from the same task they don't have issue's.

                                • 13. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                                  Bradley_M Apprentice

                                  Just got done with a long webex session with support but we think we found my problem.

                                   

                                  If you target a machine at a certain IP address and it encounters a different machine at that address it will "discover" that machine for up to 25 hours according to support.  This is the way it is coded at the moment so it is operating "as expected".

                                   

                                  This is more noticeable with large target groups because you are going to run into more of these IP mismatches which will eat away at the tasks resources.  I haven't heard engineering's response yet but will follow up to the discussion when I do.

                                   

                                  Bradley

                                  • 14. Re: Agent Discovery is taking 20+ minutes on LANDesk 9.6 SP2
                                    kleong Apprentice

                                    Bradley,

                                     

                                    We too are experiencing the same issue as you are (time it takes for endpoints to be reported as "off" and the time it takes to complete a "push" scheduled task) - we're just engaging support now about it.

                                     

                                    We had to recently push out to 4000 endpoints and it literally took over 4hrs to go through all of them - before we upgraded to LDMS v9.6 SP2 from LDMS v9.5 SP2, we did not see this problem and back with LDMS v9.5 SP2, we would have been completed within an hour or less.

                                     

                                     

                                    Kevin