5 Replies Latest reply on May 16, 2008 3:18 AM by phoffmann

    kill multicast distribution job

    Rookie

       

      I have set up a multicast distribution job - installing Office 2003, which I have now discovered is a bad idea....

       

       

      I have cancelled the Active job in Scheduled Tasks, rebooted the machine, stopped and started the tmcsrv  to no avail... the machine simply seems to pick up from where it left off and continue downloading files from the UNC share to the cache. Is there any way to stop this?

       

       

      Michael

       

       

        • 1. Re: kill multicast distribution job
          phoffmann SupportEmployee

          A few things here.

           

          1 - What version of LANDesk are you using.

          2 - What delivery method are you using. Is it a "push" or a "policy-supported-push".

          3 - Which machine are you exactly talking about in terms of having rebooted - is that the Core Server?

           

          It's not exactly "normal" behaviour for a job to continue after you've killed it. "worst case scenario" you can kill the Scheduler Service on the Core and associated TMC-services while you go about deleting the job, but it depends on what exactly you've configured and what exactly you're using, as 8.8 changes a fair few rules (at least in terms of policies) in this regard.

           

          Paul Hoffmann

          LANDesk EMEA Technical Lead.

          • 2. Re: kill multicast distribution job
            Rookie

             

            Hi Paul,

             

             

            LANdesk is version 8.7 sp4 - Push delivery -I have rebooted the client... not the server. 

             

             

            Essentially the only thing I can see happening at the moment is the files being slowly added to the cache directory and the tmcsvc.log file giving progress reports. I am not particularly worried by this behaviour as all I have read (since) indicates this is normal.... but what now that I want to kill the job. It (should) no longer be receiving instructions from the core to run it, so it must be some sort of job queue on the client, but I can't find it?

             

             

            I hope this makes sense to you. Regards,

             

             

            Michael

             

             

             

             

             

            • 3. Re: kill multicast distribution job
              phoffmann SupportEmployee

              Well - in short, SDCLIENT gets given a "job"-file, which tells it what to do.

               

              Part of this includes (in this case) the "go forth and listen to thy multicast, and pick up all of it" ... so it will. Because the job didn't complete, it's quite likely that SDCLIENT simply noticed "Oh, this is my multicast" as it was listening and - between the multicast and peer download - went about its business of getting all the files it should have.

               

              It's somewhat the point of SDCLIENT - "to reliably get files down" so to speak :).

               

              So - what is it exactly you want to ultimately acheive (I'm a bit confused on why you don't want the client to receive files) here. In theory it should be enough to delete the task-manifests and the task-status files on the client, but it'd help if you'd explain a bit better what exactly you're trying to do/acheive and why, if possible.

               

              Thanks.

               

              Paul Hoffmann

              LANDesk EMEA Technical Lead.

              • 4. Re: kill multicast distribution job
                Rookie

                 

                Hi Paul,

                 

                 

                Sorry for being so vague. I am new to LANdesk, so trying to come to terms with the way the client functions. I apologise, as my terminology might be a bit awry too. I was sort of trying to follow the "Installing Microsoft Office 2003 Best Known Methods" document along the way....

                 

                 

                Initially, I set up using a Standard Push - Method 1 as outlined in the above document - kicked it off, and it seemed to take forever.

                 

                 

                So then I read further... and decided to try a Full Speed Push - Method 3 in the BKM. But it wouldn't kick off as the first process was still in play.

                 

                 

                Normally, I would not want to stop the client receiving files. I do appreciate the concept of SDCLIENT reliably getting files down, however ... what if someone made a big mistake and tried to push some massive piece of software out to a large number of machines using the wrong method. How would you go about stopping this? I am asking out of interest now as my job eventually completed happily at about 10:30 last night anyway in spite of my feeble attempts to kill it.  I suppose I am asking exactly where to find task-manifests and task-status-files, and an overview of how they are used by the system, etc.

                 

                 

                regards,

                Michael

                 

                 

                • 5. Re: kill multicast distribution job
                  phoffmann SupportEmployee

                  Not a problem. Thanks for taking the time to explain the context - that helps a lot, and now it makes a lot more sense

                   

                  First of all, Log-files for software distribution (client-side) can be found in two places

                   

                  1 - "C:\Program Files\LANDesk\LDClient\"

                  ==> SDCLIENT.LOG

                   

                  2 - "C:\Program Files\LANDesk\LDClient\Data\"

                  => SDCLIENT.LOG

                  => SDCLIENT_TASK##.log

                   

                  In most cases, particularly the SDCLIENT_TASK##.log (where ## == the number/TASK_IDN of the task in the database's "LD_TASK" table) will be useful and tell you where the client has downloaded his files from (i.e. what peer / which preferred package server).

                   

                  This is "general log stuff" of interest to you.

                   

                  ========

                   

                  Now a few other files

                  3 - "C:\Program Files\LANDesk\LDClient\sdmcache\"  ... this holds:

                  - taskmanifest.CORESERVERNAME.TaskIDNumber.Number.ini

                   

                  This is the information for the job in terms of "download this file / these files and they have this hash / these hashes".

                   

                  ========

                   

                  4 - "C:\Program Files\LANDesk\LDClient\" ... this holds

                  - SDCLIENT.TASKS.XML

                   

                  This is the actual file that keeps track of where a job is.

                   

                  =========

                   

                  5 - "C:\Program Files\LANDesk\LDClient\Data\" ... this olds

                   

                  - TASKS_LDAPINFO.INI

                  This holds information relevant to LDAP-based tasks. This may or may not hold anything useful/relevant for you.

                   

                  That should be enough to get you started.

                   

                  As a crude approach to "understanding how stuff works" - what you can do is to run FILEMON (from SYSINTERNALS) against a directory (i.e. - LDCLIENT - or even LANDesk above it) and launch a software distribution job and see what executables are called.

                   

                  This assumes that you've disabled/stopped "high access"-processes/services like SOFTMON (Software License Monitoring) though, as those will otherwise spam you to oblivion :).

                   

                  Filemon and Procmon are pretty neat to understand these sorts of things :).

                   

                  - Paul Hoffmann

                  LANDesk EMEA Technical Lead.