locked
Concurrent connections to Distriubution Point RRS feed

  • Question

  • Hey All, SCCM newbie here,

    I've just configured SCCM2012 and is running nicely. I'm currently deploying a critical driver update to approximately 600 devices and it is taking a very long time. (about 15 per day). In checking the application deployment status it never shows more than 3 devices downloading/installing at any given time.
    I know my server and network could easily handle the extra load of say, (at least) 30 concurrent devices.
    I've snooped around the settings and have found options for concurrent transfers of software between sites and distribution points but not between distribution points and clients.

    Does such a setting exist? Surely this thing can have more than 3 concurrent clients recieving software deployments?
    Could it be anything to do with devices still showing up in Assets and Compliance and receiving the client?

    Thanks,

    Jackson

    Tuesday, April 12, 2016 10:58 PM

Answers

  • "The client polling interval is set to 5 minutes."

    That is very bad and may be the source of your issue. The client simply doesn't have enough time to catch it's breath let alone actually download and apply the machine policy. Never lower this to less than 15 minutes and is reality, never below 30 minutes. In the vast majority of deployments, the default of 60 minutes works just fine.


    Jason | http://blog.configmgrftw.com | @jasonsandys

    • Marked as answer by Jackson R Thursday, April 21, 2016 2:32 AM
    Thursday, April 14, 2016 2:36 AM

All replies

  • No, there is no explicit option in ConfigMgr for this. It could be controlled directly in IIS, but I don't generally recommend doing it there.

    BITS throttling could be effectively doing this; do you have any BTS throttling configured?

    There may be some other external factor influencing this. How big is the package? 

    I have no idea what your last question means...


    Jason | http://blog.configmgrftw.com | @jasonsandys

    Wednesday, April 13, 2016 12:22 AM
  • Thanks for the response Jason.

    I do not have any BITS throttling enabled.

    The package is approximately 700mb, I wouldn't have thought that would be an issue however I may be wrong.

    In regards to my last question, currently the clients on my network are still having the SCCM client pushed out to them as they are still being discovered. Could this be getting in the way of my deployment?

    Thanks again,

    Jackson

    Wednesday, April 13, 2016 2:06 AM
  • Well, yes, on the client agent installation. No client = not managed by ConfigMgr = deployment is meaningless. Thus, until a system gets the client agent installed and it is properly communicating with the site, there is no way for it know about let alone act upon the deployment.

    Jason | http://blog.configmgrftw.com | @jasonsandys

    Wednesday, April 13, 2016 2:13 AM
  • Thanks Jason, I'm aware that the client needs to be installed before a deployment can begin. (Sorry I worded that badly)

    What I am referring to is devices that already have the client installed.

    I have 18 devices waiting in the cue that are reporting that they are online. However no more than 3 devices are ever "in progress". I've even manually run the "Machine Policy Retrieval & Evaluation Cycle" action (yesterday) on a client but it still hasn't received anything. It remains in the "unknown" tab of Deployment Status.

    Jackson  


    • Edited by Jackson R Wednesday, April 13, 2016 2:31 AM
    Wednesday, April 13, 2016 2:29 AM
  • Do the resources for these systems show in the collection that you've targeted with the deployment?

    Are the systems getting other deployments?

    Have you reviewed the logs on the systems directly to validate that they are healthy and properly communicating with the site? ccmmessaging.log, policy evaluator.log, clientlocation.log, and locationservices.log are the ones I'd start with?

    Is the content for the deployment deployed to a DP accessible to all of the systems in question?

    Do all of the systems in question fall into a defined boundary which in turn is associated with a boundary group that is itself associated with a DP that has the content?


    Jason | http://blog.configmgrftw.com | @jasonsandys

    Wednesday, April 13, 2016 2:43 AM
  • The systems do show up in the collection that I have targeted. They are reporting that they are active.

    This is my first mass deployment since installing sccm, so apart from lab testing, I have not deployed anything on this scale. Other systems (all identical hardware) in the collection are receiving the deployment. Just very slowly. 38 devices have received the deployment in the past 2.5 days. 5 of them in the past 4 hours.

    I had a brief look at the logs you mentioned but i'm not familiar on what to look for.

    The content is deployed to a DP that is associated with a boundary group containing my entire AD.

    As stated earlier, the system looks to be working OK, clients are receiving the packages OK. It just appears to be limited to 3 threads at a time.

    Also, if it's any help, the host is running Server2012r2, 8gig ram, 280gb free hdd space, 2 x Xeon E5-2640 cores, 1gb nic.

    Networking backbone is 10gb, served to Wireless N/AC AP's.

    Clients are wireless in this case, but for testing, my lab PC is wired.

    Wednesday, April 13, 2016 3:25 AM
  • The content is deployed to a DP that is associated with a boundary group containing my entire AD.


    What is that in technical terms? Is there just a single AD site containing all Clients?

    Torsten Meringer | http://www.mssccmfaq.de

    Wednesday, April 13, 2016 5:57 AM
  • Correct. This is a single site setup with a single DP.
    Wednesday, April 13, 2016 10:20 PM
  • There are no threads per se. All Client activity in ConfigMgr is client initiated and there is nothing to limit how many can connect at one time in ConfigMgr itself.

    How many total clients are in the collection targeted by the deployment?

    Have you changed the default machine policy polling interval?

     

    Jason | http://blog.configmgrftw.com | @jasonsandys

    Thursday, April 14, 2016 1:41 AM
  • There are 688 devices in the collection, of those 688, 59 have installed the client and are reporting they are online (the rest are in storage/not seen by network etc) and are in the "unknown" tab of Deployment Status.  They are slowly making there way from Unknown to In Progress and then to Success, its just very slow. (46 are listed as successful in nearly 4 days)
    When attempting to manually run the "Machine Policy Retrieval & Evaluation Cycle" action on a test machine, after 2 days (and nights) the machine still did not receive the package. Others did, however.

    The client polling interval is set to 5 minutes.

    Thanks again for the responses.

    Jackson

    Thursday, April 14, 2016 2:21 AM
  • "The client polling interval is set to 5 minutes."

    That is very bad and may be the source of your issue. The client simply doesn't have enough time to catch it's breath let alone actually download and apply the machine policy. Never lower this to less than 15 minutes and is reality, never below 30 minutes. In the vast majority of deployments, the default of 60 minutes works just fine.


    Jason | http://blog.configmgrftw.com | @jasonsandys

    • Marked as answer by Jackson R Thursday, April 21, 2016 2:32 AM
    Thursday, April 14, 2016 2:36 AM
  • OK I didn't realise, I'll readjust and get back to you.

    Thanks!

    Thursday, April 14, 2016 2:44 AM
  • I've adjusted and haven't seen a real increase in software deployments in the last 21 hours or so. I did however notice a jump in the number of machines that had the client installed, so that's good.

    Any other suggestions? I created an additional deployment to a different collection but the total number of machines receiving deployments (across both collections) at any time remaines at 3 or 4.

    Friday, April 15, 2016 12:24 AM
  • Hi Jason,

    After monitoring the SCCM console this morning I have noticed that a much larger number of devices have begun to receive the deployment (up to 16 at a time). It seems like adjusting the polling interval (30mins) seems to have had a positive effect.

    I'll continue to monitor but so far it a clear improvement.

    Thanks,

    Jackson

    Monday, April 18, 2016 12:31 AM